Eighty-eight percent of CMOs are being asked by their boards about AI visibility. Most are trying to solve it by hiring. That is the wrong answer. AEO is a specialist function, and specialist functions belong in your agency stack.

Eighty-eight percent of CMOs are being asked by their boards about AI visibility. Most are trying to solve it by hiring. That is the wrong answer. AEO is a specialist function, and specialist functions belong in your agency stack.

The conversation in most marketing leadership teams right now goes something like this: the board asks about AI visibility, the CMO commits to a plan, and someone on the team suggests hiring an AEO specialist. It sounds reasonable. It is almost always the wrong move.

The instinct to hire is understandable. When a new capability becomes strategically important, the reflex is to bring it in-house. That is how most marketing teams built their SEO, paid media, and content functions over the past two decades. But AEO is not at that stage of maturity, and the conditions that made in-house SEO sensible do not apply to in-house AEO in 2026.

<h2>The Board Is Already Asking</h2>

The pressure is real. Corporate Ink's 2026 GEO and AI Visibility Report found that 88% of CMOs and VP-level marketers are being asked by their boards or leadership about what they are doing to optimize for AI visibility. [1] That is not a future problem. It is a present one. And the gap between board expectation and marketing team capability is wide: only 26% of marketing teams know which media outlets AI engines actually crawl in their market, and only 17% have earned coverage in those outlets in the past month.

<img src="/manus-storage/cmo_pressure_generated_2d3dad7b.webp" alt="The Board Is Already Asking" class="w-full rounded-lg my-6" />

The same research found that 72% of brands are being described inaccurately by AI engines right now. A buyer who asks ChatGPT or Perplexity about a vendor and receives a description that is wrong, whether it includes an old product, a mischaracterized differentiator, or positions the company in the wrong category, may never make it to the company's website to correct that impression. This is an active pipeline risk, and it is happening at scale.

<h2>Why In-House AEO Fails the Build Test</h2>

Building genuine AEO capability in-house requires a minimum of three specialized roles: an AEO strategist, a technical SEO engineer with structured data expertise, and a content specialist trained in entity-dense, citation-optimized writing. The minimum annual cost before a single article is published or a single schema tag is deployed is $280,000 to $450,000. [2] That figure does not include the $2,000 to $5,000 per month in tool licensing for AI citation monitoring platforms, multi-model testing environments, and competitive intelligence tools.

The talent market makes the cost problem worse. Fewer than 2,000 professionals globally have genuine multi-model AEO experience as of mid-2026. Most people calling themselves AEO specialists are rebranded SEO practitioners who lack hands-on experience with citation rate measurement, entity graph orchestration, or cross-platform authority engineering. Recruiting a qualified AEO strategist takes an average of 4.5 months. During those months, competitors continue compounding their advantage.

The timeline problem is the most underappreciated. An in-house team building its first AEO playbook requires 12 to 18 months before producing measurable citation gains. An agency that has optimized dozens of brands can deploy proven playbooks in 90 days. That asymmetry is not incremental. It is categorical. And in a market where AI citation authority compounds over time, the 12-month delay is not a cost. It is a structural disadvantage.

<img src="/manus-storage/build_vs_buy_generated_2cb3759c.webp" alt="Build vs Buy: The True Cost of AEO Capability" class="w-full rounded-lg my-6" />

Digital Strategy Force's analysis of the AEO services market found that fewer than 5% of organizations should build in-house, and that profile requires five criteria to be met simultaneously: $2M or more in AI search revenue exposure, an 18-month timeline tolerance, an executive AEO sponsor with budget authority, FTE hiring authority for three specialized roles, and a proprietary data asset that justifies the investment. Most brands meet two or three of those criteria. That is not enough.

<h2>The Agency Stack Model</h2>

The right mental model for AEO is the one that already governs every other specialist marketing function. The CMO does not run paid media campaigns internally. The CMO does not write press releases and pitch journalists. The CMO does not manage programmatic ad buying or affiliate networks. These functions are specialist disciplines with their own tools, methodologies, talent markets, and measurement frameworks. They belong in the agency stack because that is where specialist capability lives.

AEO is no different. It requires a specialist understanding of how AI engines evaluate trust signals, how entity graphs work, how citation authority compounds across platforms, and how to measure share of voice in an environment where there are no rankings to track and no clicks to count. That knowledge does not exist inside most marketing teams, and it cannot be acquired quickly by promoting an SEO manager.

<img src="/manus-storage/agency_stack_generated_82cf4dbc.webp" alt="The CMO Agency Stack Model" class="w-full rounded-lg my-6" />

The CMO's job in this model is not to execute AEO. It is to own the strategy: define the brand's AI visibility goals, set the measurement framework, brief the agency on the competitive landscape and the buyer questions that matter most, and hold the agency accountable for citation rate improvement. That is the same relationship a CMO has with a PR agency, a paid media agency, or an SEO agency. The division of labour is not a weakness. It is how specialist functions work.

<h2>What the Right AEO Agency Actually Does</h2>

A genuine AEO agency is not an SEO agency that has added AI to its service menu. The work is categorically different. Where SEO optimizes your own pages to rank, AEO optimizes the entire ecosystem of signals that AI engines use to evaluate your brand's claims. That includes your own content, but it also includes the third-party sources that describe you, the entity signals that anchor your brand to a knowledge graph, and the answer architecture of every piece of content your brand publishes.

Corporate Ink's research found that among companies experiencing pipeline growth from AI visibility, 55% say their PR or AEO agency is prioritizing AI visibility and actively integrating it into their strategy. Among companies not seeing pipeline growth, half say their agency has raised the topic but does not have the expertise to make a meaningful difference. [1] The agency you choose matters as much as the decision to use one.

The signals that distinguish a genuine AEO agency from a rebranded SEO shop are specific. A real AEO agency tracks citation rates across multiple AI platforms, not just Google AI Overviews. It has a methodology for entity management, not just keyword research. It can audit what AI engines are saying about your brand and identify the specific third-party sources that need to change. It measures share of voice in AI answers, not just organic rankings. And it can connect AEO investment to pipeline outcomes, not just visibility metrics.

<h2>The Prompt Group Model</h2>

The agency model that makes the most sense for AEO is one that combines the deep technical expertise of a specialist firm with the strategic perspective of a brand advisor. <a href="https://www.thepromptgroup.com" target="_blank" rel="noopener noreferrer" class="text-amber-700 underline">The Prompt Group</a> was built specifically for this. It is not an SEO agency that has added AI visibility to its service list. It was founded on the premise that AI citation authority is a distinct discipline that requires a distinct methodology, and that most brands will get there faster and more reliably by working with a specialist partner than by building the capability from scratch.

The practical starting point for most CMOs is an AI visibility audit: a structured assessment of what AI engines are currently saying about your brand, which queries your brand appears in and which it does not, where the entity signals are weak, and which third-party sources need to carry different claims. That audit takes two to three weeks with a specialist agency. It takes six to nine months for an in-house team to develop the methodology to do it at all.

<h2>The Strategic Question</h2>

The board is not asking whether your brand has an AEO hire. It is asking whether your brand is visible in AI answers. Those are different questions, and confusing them leads to the wrong decision. Hiring an AEO specialist is a means. AI visibility is the end. The fastest, most reliable path to the end is a specialist agency with a proven methodology, cross-industry pattern recognition, and the measurement infrastructure to prove it is working.

The CMO who treats AEO as an in-house build project will spend 18 months and $400,000 discovering what a specialist agency already knows. The CMO who adds AEO to the agency stack will have measurable citation gains in 90 days and a board answer that is grounded in data rather than aspiration.

The agency stack model is not a concession. It is the strategy.

<h3>References</h3>

AI Overviews appear in 68% of local searches. Local packs appear in 39%. That 29-point gap is where local businesses are losing customers they never knew they had.

AI Overviews appear in 68% of local searches. Local packs appear in 39%. That 29-point gap is where local businesses are losing customers they never knew they had.

There is a number worth sitting with before you read any further. In a study of 540 local queries across six industries and three major US cities, Whitespark found that AI Overviews appeared in an average of 68% of local searches. Local packs appeared in 39%. [1] That is a 29-point gap between the two most important local search surfaces, and almost no local business is optimized for both.

The gap is not a rounding error. It is a structural split in how Google now handles local intent, and it has a direct consequence for every business that has spent the last decade building a local SEO program: you can rank first in the map pack and be completely invisible in the AI answer for the same query. That is not a ranking problem. It is a category problem.

<h2>Why the Two Systems Run on Different Signals</h2>

Local SEO and local AEO are not the same discipline with different names. They are optimized for different systems that evaluate different signals and produce different outputs.

The local pack is a proximity and relevance engine. Google Business Profile completeness, review volume and recency, citation consistency, and physical proximity to the searcher are the primary inputs. The output is a ranked list of nearby businesses. The optimization target is clear: be close, be complete, be reviewed.

AI Overviews are an authority and informativeness engine. The Whitespark study found that AI Overviews dominate informational and hybrid queries, appearing in 92% of informational searches and 97% of hybrid searches, while local packs appear in only 6% and 17% of those same query types respectively. [1] The AI is not asking which business is closest. It is asking which source is most credible for this question.

This is the structural split. A business that has invested entirely in GBP optimization, citation building, and review management has built a strong local pack presence. But if a potential customer types a hybrid query like 'best dentist for kids in Phoenix' or 'how much does a personal injury lawyer charge in Houston', the AI Overview is the dominant result and the local pack may not appear at all.

<h2>The 29-Point Gap in Practice</h2>

The Whitespark data breaks down by query intent in a way that makes the strategic implication concrete. For traditional local-intent queries like 'plumbers in Denver', local packs appear 93% of the time and AI Overviews appear only 15% of the time. Local SEO is still the right tool for that query type. But for the informational and hybrid queries that represent a growing share of local search volume, the picture inverts completely.

A BrightLocal survey published in 2026 found that 45% of consumers now use AI tools like ChatGPT, Gemini, or Perplexity for local business recommendations, up from 6% one year earlier. [2] That is not a slow trend. That is a behavioral shift happening faster than most local marketing budgets can respond to.

The OmniEclipse AI Search Visibility Report found that only 11.9% of businesses appear in AI search results, meaning 88% are completely absent from AI-driven discovery. [3] Cross that with the Whitespark finding that AI Overviews appear in 68% of local searches, and the math becomes uncomfortable: most local businesses are invisible in the search surface that now dominates most of the queries their potential customers are running.

<h2>What Local AEO Actually Requires</h2>

The signals that drive AI Overview inclusion are not the same as the signals that drive local pack rankings. Whitespark's own analysis found that 60% of AI Overview citations in local searches point to third-party publishers, including sites like Yelp, Reddit, HomeGuide, Thumbtack, and Quora. Only 40% cite individual business websites directly. [1] That ratio has a strategic implication: being cited in the right third-party directories and review platforms is not just a citation-building tactic. It is an AI visibility tactic.

The content signals matter too. AI Overviews for hybrid queries are pulling from FAQ content, blog posts, and informational pages that answer the questions customers are actually asking before they decide to buy. A plumbing business that has published a clear, well-structured page answering 'how much does it cost to replace a water heater in Denver' is a candidate for AI Overview inclusion on that query. A business with only a GBP listing and a homepage is not.

The Whitespark AI Mode guide identifies four practical priorities for local businesses: tighten NAP accuracy across all platforms, document customer FAQs in natural conversational language, audit which sources AI Mode cites for your industry and top competitors, and double down on reviews and unstructured citations from blogs, news sites, and community pages. [4] None of those are new tactics in isolation. What is new is the reason for doing them: not to rank in the map pack, but to be cited in the AI answer.

<h2>The Measurement Problem</h2>

One of the harder aspects of local AEO is that AI Mode results are not consistent across users. Whitespark documented that two people making the same search from the same location can see completely different AI Mode outputs. [4] This makes traditional rank tracking tools inadequate for measuring local AI visibility. You cannot track a single position because there is no single position to track.

The practical response is to shift from position tracking to citation tracking. Run a fixed set of hybrid and informational queries in your category and geography weekly. Note which businesses, which third-party sources, and which content types appear in the AI answers. Track whether your brand, your website, or your third-party listings are cited. That is your local domain territory, and it is the leading indicator that matters.

<h2>What to Do This Week</h2>

The businesses that will close the 29-point gap are the ones that treat local AEO as a parallel discipline rather than a replacement for local SEO. The map pack is not going away. Local-intent queries still return local packs 93% of the time. But the informational and hybrid queries that precede a purchase decision are increasingly being answered by AI, and those answers are being built from a different set of signals than the ones your current local SEO program is optimizing for.

Run your ten most important local queries this week. For each one, note whether a local pack appears, whether an AI Overview appears, and whether your business is cited in either. The gap between those two counts is your local AEO opportunity. For most local businesses, that gap is larger than they expect.

<h3>References</h3>

ChatGPT just shipped its version of Google's not provided. The organic attribution you keep asking for is not coming. Here is the measurement stack you can build today without permission from any platform.

ChatGPT just shipped its version of Google's not provided. The organic attribution you keep asking for is not coming. Here is the measurement stack you can build today without permission from any platform.

There is a conversation happening in every AEO strategy meeting right now, and it goes roughly like this: we know AI visibility matters, but we cannot prove the ROI, so we cannot get the budget. The implication is that the proof is coming, that some future version of ChatGPT Analytics will hand us a clean line from AI citation to revenue, and then we can finally make the case. That conversation is based on a false premise, and the sooner you let it go, the faster you can build a measurement stack that actually works.

Duane Forrester, writing in Search Engine Journal this week, made the structural argument clearly: ChatGPT has already built a full closed-loop conversion attribution system. It has a pixel, an events API, standard e-commerce events, deduplication between browser and server, and a privacy-preserving identifier that ties ad exposure to purchase. That system exists right now. It sits behind an ad account. Organic practitioners get a robots.txt file and best wishes.

This is not an accident or an oversight. It is the same playbook Google ran in 2011 when it encrypted organic search referrals and keyword data vanished into the not-provided bucket. Paid search advertisers kept getting richer and richer conversion data. Organic SEO spent a decade angry about it. ChatGPT did not even need to take anything away. It simply never granted organic attribution in the first place. A designed absence generates no protest, only a low, unfocused unease. That knowledge was ambient. Any competent operator building an answer engine already had it.

The commercially driven platforms, OpenAI, Perplexity, Google, will build measurement for the people spending money. The mission-driven ones, Anthropic, have better things to build. None of them has an incentive to hand free, item-level organic attribution to practitioners. Do not expect it from ChatGPT. The real version of that statement is: do not expect it anywhere.

<h3>Attribution Is Actually Three Different Problems</h3>

When practitioners say attribution, they are usually blending three problems that have three different answers. Understanding the distinction is the first step to building a measurement stack that does not depend on platform cooperation.

The first problem is referral attribution: did an AI answer link to you, did someone click, did that session convert. This is measurable today. The click carries a referrer into your own analytics. The catch is that AI referral traffic is currently under one percent of total traffic on most sites, which means the number you get is real but small. Treat it as a floor, not a ceiling, because unattributable and cross-window purchases bias every honest count downward.

The second problem is incrementality: not who clicked, but whether your visibility caused lift you would not otherwise have gotten. This is measurable too, but it requires design rather than a dashboard. Hold out a set of geographies and change nothing in them while you push AEO work everywhere else. Run on-off tests over defined windows. Track a fixed set of queries before and after a content push and watch what moves. This is causal measurement, and it belongs to you, not to the platform.

The third problem is influence, the dark funnel case: the buyer read your brand inside an AI answer, never clicked, and showed up three weeks later through a branded search. This is genuinely hard to measure, but it is not unmeasurable. Self-reported attribution, the how-did-you-hear-about-us field at the point of conversion, catches influence that no pixel will ever see. Branded search lift correlation, watching whether branded search and direct navigation rise as your AI visibility rises, gives you a defensible read at the aggregate level.

<h3>Domain Territory Is Your Leading Indicator</h3>

The measurement frame that works for AEO is not the one borrowed from paid search. It is closer to the one used for brand equity measurement: you track the territory you own in the conversation, and you watch whether that territory expands or contracts over time.

Domain territory, the share of AI-generated answers in your category that include your brand, your claims, or your framing, is a leading indicator of commercial outcomes. It is not a lagging indicator like revenue attribution. It tells you whether you are building the kind of authority that produces citations before those citations produce clicks, and long before those clicks produce conversions.

The practical measurement stack looks like this. Track your brand mention rate across a fixed set of category queries in ChatGPT, Perplexity, Claude, and Google AI Mode. Do this weekly. Track the specific claims and framings that appear alongside your brand mentions. Track whether your competitors are gaining or losing territory in the same queries. Track branded search volume in Google Search Console as a proxy for dark-funnel influence. None of this requires permission from any platform.

<h3>The Vendors Worth Trusting</h3>

The AI visibility measurement space is filling up with vendors making claims that do not survive scrutiny. The question that separates credible from theatrical is simple: what data source closes the loop? If the honest answer is first-party data, their agents on your site, your analytics, your CRM, then what they sell is real but bounded. It lives at the referral layer. If the answer implies signal drawn from inside OpenAI or Anthropic, they are either misrepresenting a referrer-detection method or lying to you. There is no third source.

The vendors doing this credibly are the ones who sidestep the black box entirely and run on first-party data. They compute attribution by joining their own query logs to your analytics, not by claiming access to platform internals. Ask the question. Listen carefully to the answer.

<h3>What to Tell the CFO</h3>

The budget conversation changes when you stop asking for click attribution and start presenting domain territory as a strategic asset. The argument is not that AEO produced X conversions last quarter. The argument is that your brand is cited in Y percent of AI answers in your category, up from Z percent six months ago, and that branded search volume has risen in parallel. The causal chain is probabilistic, not deterministic, but it is the same chain that justifies brand advertising, sponsorship, and thought leadership investment. Every CMO already funds activities where the attribution is modeled rather than counted.

The measurement world you are standing in was already modeled, probabilistic, and permission-dependent before answer engines entered the picture. The deterministic era was ending before a single LLM shipped. Third-party cookies, Apple's App Tracking Transparency, GA4's modeled conversions: all of it frayed the clean line before ChatGPT arrived. LLMs did not break attribution. They arrived after the break and made it impossible to keep pretending.

Stop waiting for a clean line that no longer exists. Build the measurement stack you can build today. Track domain territory. Run incrementality experiments. Watch branded search. Report attribution as attribution and reserve the word incremental for the cases where you ran the test. That discipline, applied consistently, will produce more credible evidence than any platform attribution system you are waiting for.

The White House is debating whether to restrict Chinese open-weight AI models on national security grounds. The All-In Podcast, David Sacks, and Jensen Huang all agree: a ban would be an economic own goal that hands China exactly what it wants.

The White House is debating whether to restrict Chinese open-weight AI models on national security grounds. The All-In Podcast, David Sacks, and Jensen Huang all agree: a ban would be an economic own goal that hands China exactly what it wants.

The Trump administration is actively debating whether to restrict American enterprises from using Chinese open-weight AI models. The proximate trigger is Kimi K3, released in mid-July 2026 by China's Moonshot AI. It performs on par with GPT-5.6 and Claude Opus 4.8 at roughly half the cost. The White House is reportedly considering adding Chinese AI firms to the entity list and issuing an executive order making US companies liable for any security breaches from hosting Chinese models.

The argument for restriction, pushed hardest by Anthropic and to a lesser extent OpenAI, is that Chinese labs are engaged in industrial-scale IP theft by sending synthetic queries to US frontier models and using the outputs to train their own. Dean Ball, OpenAI's head of strategic futures, made the case bluntly on X: you do not need to ban open source outright. You just need to direct every agency to issue soft law that creates enough fear, uncertainty, and doubt to make every regulated enterprise back off.

That argument got torched from multiple directions simultaneously.

David Sacks, Trump's former AI czar, called it out directly: the weaponization of regulatory uncertainty as a competitive tool should be completely unacceptable. He accused the closed-lab duopoly of wanting the government to eliminate their open-source competition and called on Silicon Valley to defend open competition. Jensen Huang was equally blunt: if everything becomes one single model, one single point of failure, the world is much more vulnerable. He called Chinese open-source models excellent and dismissed the backdoor fears as a misconception.

The All-In Podcast panel on July 24 went further. Chamath Palihapitiya laid out the economic consequence with precision: if US companies are forced to pay 50 to 100 times more per token than international competitors who can freely use Chinese open-source models, US companies lose margin and the market re-rates them downward. He was direct about the stock market implication: if the US government intervenes, it will tank the stock market. Period. Not debatable.

Jason Calacanis added the startup evidence. Companies like Lovable and ElevenLabs have already moved off frontier APIs to open-source models running on their own hardware, saving 50 to 90 percent while achieving comparable quality for 95 percent of tasks. Thinking Machines, the model company led by Mira Murati, former CTO of OpenAI, was bootstrapped via distillation from a Chinese open-source model. Cursor's Composer 2 used post-training on top of a Chinese model. A ban would not hurt China. It would burn down the US startup ecosystem.

The IP theft framing also has a hypocrisy problem that the All-In hosts dissected at length. Anthropic and OpenAI have spent years arguing in court that training on all public internet content is fair use. They are now calling the same practice, applied to their own outputs, industrial-scale theft. Friedberg made the logical trap explicit: if Anthropic wins the distillation-equals-theft argument in Washington, it implicitly concedes the same argument to the New York Times, book authors, and music labels suing them. The $1.5 billion copyright settlement Anthropic agreed to on July 20, 2026 suggests they know their fair-use case is weaker than they claimed.

The correct policy response, as Sacks argued, is to ban Chinese labs from querying American frontier models, not to ban American companies from using Chinese ones. You are punishing the wrong side. The source of the distillation problem is American models being accessible to Chinese labs. Fix that. Do not cripple American developers to protect the pricing power of two incumbents.

For AEO practitioners, this debate has a direct operational dimension. If the White House restricts Chinese open-weight models, the model routing pool shrinks. Brands that have built citation visibility in DeepSeek, Kimi, or Qwen lose that surface area overnight. More importantly, the smaller, cheaper models that routers increasingly prefer for cost reasons are disproportionately Chinese open-weight. A ban would push routers back toward expensive frontier models, raising the cost of every AI query and slowing the adoption curve that is driving AEO's growth as a discipline. The brands that want AEO to matter should be rooting for open-weight models to thrive, not for Washington to protect a duopoly.

Satya Nadella confirmed that Microsoft own MAI models are now handling tasks in GitHub Copilot, Excel, and Outlook that used to go to GPT-4. For brands optimizing for AI citations, the model behind the answer just changed.

Satya Nadella confirmed that Microsoft own MAI models are now handling tasks in GitHub Copilot, Excel, and Outlook that used to go to GPT-4. For brands optimizing for AI citations, the model behind the answer just changed.

Microsoft has started routing tasks inside GitHub Copilot, Excel, and Outlook to its own in-house MAI models, replacing work that previously went to OpenAI and Anthropic. CEO Satya Nadella confirmed the shift in a post on X, describing it as a cost and capability decision. Copilot Chat and PowerPoint are next in line.

The MAI models in question include MAI-Image-2.5-Pro and MAI-Voice-2-Flash, which Microsoft says cut GPU costs by up to 89 percent compared to frontier alternatives. Nadella framing is direct: software now has a real marginal cost for the first time, which makes the choice of model a business decision rather than a technical default. Frontier models stay in the mix for frontier problems. Everything else goes to whatever is cheapest and good enough.

For enterprise users, this is largely invisible. Copilot still answers your questions. The model behind the answer is just different. That invisibility is precisely the problem for AEO practitioners.

AEO strategies built around ChatGPT citation patterns assume that Copilot behaves similarly, because both historically ran on GPT-4. That assumption is now wrong for a growing share of Copilot queries. Microsoft MAI models were trained on different data, with different fine-tuning priorities and different retrieval behaviors. A brand that has built strong citation visibility in ChatGPT may not have the same visibility in a Copilot session that routes to MAI-Image or a Phi-4 variant.

The enterprise context makes this particularly significant. Copilot is the AI interface for most Fortune 500 employees. It handles procurement research, vendor comparisons, market analysis, and competitive intelligence queries. These are exactly the high-intent, high-value queries where brand citation matters most. If Microsoft own models are now answering those queries, and if those models have different citation patterns than GPT-4, then enterprise AEO strategies need to be recalibrated.

The practical implication is the same as the OpenRouter story: the model behind the answer is no longer a constant. Microsoft has made it a variable it controls. Brands doing AEO need to test citation visibility in Copilot directly, not infer it from ChatGPT performance. The two products are diverging, and the gap will widen as Microsoft pushes more tasks to its own models.

None of this means GPT-4 is going away from Copilot. Nadella was explicit that frontier models stay for frontier problems. But the definition of what counts as a frontier problem will keep shrinking as MAI models improve. The direction of travel is clear: Microsoft wants to own the model layer inside its own products. For brands, that means the citation landscape inside the most widely deployed enterprise AI tool just became harder to predict and more important to monitor.

Stripe's reported $10B move on OpenRouter is not a payments story. It is an infrastructure story. And for brands doing AEO, the rise of model routing changes who sees your content and when.

Stripe's reported $10B move on OpenRouter is not a payments story. It is an infrastructure story. And for brands doing AEO, the rise of model routing changes who sees your content and when.

Stripe is in talks to acquire OpenRouter for approximately $10 billion, according to the Wall Street Journal. The deal would value the AI model marketplace at nearly eight times its $1.3 billion valuation from just two months ago. That gap tells you everything about how fast the market has decided that model routing is critical infrastructure.

OpenRouter is not a model. It is a switchboard. It gives developers a single API endpoint that routes queries to whichever underlying model is cheapest, fastest, or most capable for a given task. Cursor built routing directly into its code editor. Ramp is building a routing product for enterprise token spend. Databricks has routing capabilities. The category went from niche to crowded in about six months.

Axios framed the deal well: Stripe is not buying an AI company. It is buying the metering and billing layer for inference. As tokens become the new currency of the internet, Stripe wants to sit in the middle of every transaction. OpenRouter CEO Alex Atallah has previously compared his company to Stripe. Now Stripe apparently agrees.

For AEO practitioners, the implications are direct. Model routing means that a single user query may be answered by a different model depending on cost, latency, and task complexity. A question routed to GPT-4o gets one citation pattern. The same question routed to Claude 3.5 Sonnet gets another. A question routed to a smaller, cheaper model like Mistral or Phi-4 may not surface your brand at all if your content is not in its training data.

The practical consequence is that AEO is no longer a single-model optimization problem. Brands that have built citation visibility in ChatGPT need to ask whether that visibility transfers to every model in the routing pool. The answer, in most cases, is no. Common Crawl coverage, training cutoffs, and fine-tuning decisions vary significantly across models. A brand that ranks well in GPT-4o citations may be invisible in the smaller models that routers increasingly prefer for cost reasons.

Jensen Huang put it plainly this week: if everything becomes one single model, one single point of failure, the world is much more vulnerable. He was arguing against the Anthropic and OpenAI push to restrict Chinese open-weight models. But the same logic applies to AEO strategy. Brands that optimize for one model are building on a single point of failure. The routing layer is making that risk structural.

The action for AEO teams is straightforward in principle and hard in practice: audit your citation visibility across the full routing pool, not just the flagship models. That means testing Mistral, Phi-4, Llama 3, and Gemini Flash alongside GPT-4o and Claude. It means publishing content that is structured for retrieval across models with different training data and context windows. And it means watching the routing market closely, because whoever controls the switchboard increasingly controls which brands get seen.

Most open-weight models are coming out of China. Apple has 2.5 billion devices, the best inference hardware on the planet, and a trust advantage no cloud company can buy. The question is not whether it can. The question is why it hasn't.

Most open-weight models are coming out of China. Apple has 2.5 billion devices, the best inference hardware on the planet, and a trust advantage no cloud company can buy. The question is not whether it can. The question is why it hasn't.

This argument started on a podcast. My co-host Steve and I were working through Apple's strategic position in AI, and somewhere between the Vision Pro jokes and the Siri roast, Steve said something that stuck: most open-weight models are coming out of China, and Apple is sitting on the hardware and trust infrastructure to change that. I have been thinking about it since.

The case is not complicated. It is actually obvious once you see it. But obvious things have a way of being ignored by the companies best positioned to act on them, so let me lay it out.


THE OPEN-WEIGHT GAP NOBODY IS TALKING ABOUT


When people discuss the open-weight AI landscape, they are mostly talking about Meta's Llama series, Mistral from France, DeepSeek and Qwen from China, and a handful of smaller research releases. The United States, for all its AI investment, has produced almost no serious open-weight frontier models. The dominant open-weight models that enterprises are actually running locally today are either Chinese-origin or European.

This is a geopolitical and commercial problem that nobody in the closed-model camp wants to acknowledge, because acknowledging it would require them to admit that the open-weight ecosystem is real, growing, and strategically important. Dario Amodei at Anthropic has been particularly vocal about restricting open-weight models, testifying in Washington about the dangers of releasing model weights publicly and lobbying for regulatory frameworks that would effectively make open-weight frontier models illegal to distribute.

I have written about this before, but it bears repeating here: that position is not a safety argument. It is a market protection argument dressed up as a safety argument. Anthropic already restricts genomics companies from using its models because of theoretical bioweapon concerns. Those companies are not stopping their work. They are moving to DeepSeek. The effect of Anthropic's restrictions is not less AI use in sensitive domains. It is more AI use in sensitive domains on Chinese-origin models. That is the actual outcome of the closed-model safety posture.


WHAT APPLE HAS THAT NOBODY ELSE DOES


Apple has three things that no other company can replicate.

The first is hardware. The Mac Studio with Apple Silicon is, right now, the best inference machine for running open-weight models locally. The unified memory architecture means a Mac Studio with 192GB of unified memory can run 70-billion-parameter models at speeds that would require a rack of GPUs in a data center. Apple built this hardware for its own chip roadmap and for creative professionals. The AI inference use case landed in its lap. Developers and enterprises who want to run models locally without sending data to the cloud are already buying Mac Studios for exactly this purpose. Apple is the accidental king of local inference and has not noticed.

The second is trust. Apple has spent thirty years building a brand identity around privacy. On-device processing, end-to-end encryption, differential privacy in analytics — these are not marketing slogans. They are deeply embedded in how Apple builds products. In a world where enterprises are increasingly nervous about sending sensitive data to cloud AI providers, Apple's privacy brand is a genuine competitive asset. No hyperscaler can buy that reputation. Microsoft, Google, and Amazon are all cloud-first companies whose business models depend on data flowing through their infrastructure. Apple's model is the opposite.

The third is distribution. Apple has over 2.5 billion active devices. It has direct relationships with every developer who ships on iOS and macOS. It has the App Store, the developer tools, the enterprise device management infrastructure, and the consumer trust to distribute an open-weight model to more endpoints than any research lab or cloud provider could reach in a decade.


THE SIRI PROBLEM IS A SYMPTOM


Siri is a disaster. Steve called it the Clippy of AI on our podcast, and that is being generous. Jar Jar Binks was the other comparison that came up. The point is not that Siri is bad at answering questions. The point is that Apple has been trying to build a closed, proprietary AI assistant for fifteen years and has consistently failed to make it competitive with open alternatives.

The reason is structural. Apple's AI strategy has been to build everything internally, keep it on-device for privacy reasons, and ship it as a product feature rather than a platform. That approach worked for hardware. It has not worked for intelligence. Intelligence at the frontier requires the kind of open research, public benchmarking, and community iteration that Apple's closed culture is allergic to.

An open-weight model changes the dynamic entirely. Apple does not need to win the intelligence race internally. It needs to provide the infrastructure on which the best open-weight models run, and make Apple Silicon the default substrate for local AI inference. That is a hardware and platform play, which is exactly what Apple is good at.


THE ENTERPRISE OPPORTUNITY IS ENORMOUS


Steve made a point on the podcast that I think is underappreciated: enterprises are moving back to on-premise. The cloud-first decade is over for a meaningful segment of the market. Not because cloud is bad, but because data sovereignty concerns, regulatory requirements, and the specific sensitivity of AI workloads are pushing procurement teams to ask whether they really want their internal documents, customer data, and strategic queries flowing through a third-party cloud provider.

A Mac Studio running a locally-hosted open-weight model answers that question cleanly. The data never leaves the building. The model is auditable. The inference costs are fixed and predictable. For a law firm, a hospital, a financial institution, or any company operating in a regulated industry, that is not a nice-to-have. It is a requirement.

Apple is not currently selling into this market in any meaningful way. It is not positioning Mac Studio as an AI workstation. It is not building the developer tools that would make it easy to deploy and manage open-weight models across a fleet of Apple devices. It is not even talking about this opportunity publicly. That is a gap that a competitor will eventually fill, and the most likely candidates are either a Chinese hardware company or a cloud provider trying to build an on-premise offering. Neither of those outcomes is good for Apple.


THE CONTEXT LAYER IS THE REAL PRIZE


Here is the deeper strategic argument. Apple's real competitive moat is not the iPhone. It is the context layer. Apple knows more about its users than any other company because it sits at the intersection of every device they use: the phone that tracks their location and communications, the watch that monitors their health, the laptop that holds their documents and creative work, the TV that knows their entertainment preferences.

That context layer is extraordinarily valuable for AI. An AI model that has access to your full Apple context — your health data from Apple Watch, your messages and emails, your calendar, your documents, your photos — can provide a quality of personalized assistance that no cloud AI can match, because no cloud AI has that data.

But Apple can only monetize that context layer if it keeps the data on-device. The moment it sends that data to a cloud for inference, it loses the privacy advantage and opens itself to the same regulatory and reputational risks that every other cloud AI company faces. An open-weight model running locally on Apple Silicon is the only architecture that lets Apple use its context layer at full value.


WHAT SHOULD ACTUALLY HAPPEN


Apple should release an open-weight model family under an Apple Research license. Not a product. Not a feature. A model family, like Llama, that developers and enterprises can download, fine-tune, and deploy on Apple Silicon.

It should optimize that model family specifically for Apple's unified memory architecture, so that running it on a Mac Studio or a future Apple AI workstation is meaningfully faster and more efficient than running it on commodity hardware. That creates a hardware pull-through that no other open-weight release has.

It should build first-class developer tools for model deployment and management on macOS, integrated with Xcode and the existing Apple developer ecosystem. Make it as easy to deploy a local model as it is to add a framework dependency.

And it should position the Mac Studio explicitly as the enterprise AI workstation for organizations that cannot or will not send data to the cloud. That is a real market, it is growing, and right now nobody is serving it well.


THE AEO ANGLE


For brands and marketers reading this through an AEO lens, the Apple open-weight scenario has a specific implication. If Apple ships a serious open-weight model and it becomes the default inference layer for enterprise and developer use cases, the citation and retrieval behavior of that model will matter for brand visibility in a way that the current closed-model ecosystem does not.

Open-weight models are trained on different data mixes, fine-tuned differently by different deployers, and updated on different schedules than closed models. A brand that is well-optimized for ChatGPT's retrieval patterns may be invisible to a locally-deployed Apple model that was fine-tuned on a different corpus. The AEO playbook for open-weight models is not the same as the playbook for closed models, and the gap between them will grow as open-weight deployment scales.

The practical response is the same one I have been recommending for the broader open-weight blind spot: build your brand presence in the sources that open-weight models are trained on. Common Crawl, Wikipedia, academic citations, structured data in open repositories. These are the inputs that matter for open-weight visibility, and they are systematically underweighted in most AEO strategies that were built around optimizing for GPT-4 and Claude.


THE BIGGER POINT


Steve's line from the podcast has stayed with me: the future is not about the most valued company. It is about the most trusted one. Apple is currently the most valued company in the world. It is also, by most measures, the most trusted technology brand in the world. Those two things do not always go together, and the AI transition is going to test whether Apple can convert its trust advantage into an AI leadership position before the window closes.

Building the first great North American open-weight model would be the most Apple thing Apple could do. It would be on-device, privacy-preserving, hardware-differentiated, and developer-friendly. It would be the opposite of what Anthropic is doing. And it would be the kind of move that Tim Cook's Apple has consistently failed to make because it requires betting on openness rather than control.

I think they should make that bet. I think the market is waiting for someone to make it. And I think Apple is the only North American company with the hardware, the trust, and the distribution to pull it off.

This argument started on a podcast. You can listen to the full episode here: [Unsolicited Biz Advice — Open Always Wins](https://youtu.be/6Yl3EvPqPgY?si=_Ae0VT6mphFNkvqF).

ChatGPT scaled back its checkout ambitions. Google quietly built the most complete commerce infrastructure in AI. Perplexity charges merchants nothing. Here is the honest state of LLM-native shopping and what brands should do right now.

ChatGPT scaled back its checkout ambitions. Google quietly built the most complete commerce infrastructure in AI. Perplexity charges merchants nothing. Here is the honest state of LLM-native shopping and what brands should do right now.

Somewhere between 33% and 83% of shoppers used AI to help with their holiday shopping in 2025, depending on which survey you trust. The range tells you something useful: the behavior is real and growing, but the measurement is still messy. What is not messy is the direction. AI-assisted product discovery is becoming a primary channel, and the question of whether to transact inside the AI interface — rather than clicking through to a merchant site — is the defining commerce question of 2026.

The answer, as of today, is more complicated than the headlines suggest. ChatGPT launched Instant Checkout with enormous fanfare in September 2025. By March 2026, it had scaled back. Perplexity charges merchants nothing and has built a working hybrid checkout. Google has quietly assembled the most complete commerce infrastructure of any AI platform, backed by a coalition of every major retailer and payment network on the planet. And Amazon, which invested $15 billion in OpenAI, has blocked ChatGPT crawlers from accessing its product pages.

This is not a story about one winner. It is a story about four distinct platforms, each with different economics, different audiences, and different bets on how AI commerce actually works.


THE PLATFORM LANDSCAPE


The four platforms merchants need to understand are ChatGPT, Perplexity, Google AI Mode and Gemini, and Amazon. Each operates differently enough that treating them as a single channel is a strategic mistake.

**ChatGPT** has the largest user base of any AI platform, with 810 million daily active users as of November 2025. Its Agentic Commerce Protocol (ACP), co-developed with Stripe, is an open standard that lets AI agents complete purchases on behalf of users. The vision was compelling: a shopper asks a question, gets a recommendation, and buys without leaving the chat. The reality was that only around 12 of Shopify's millions of merchants ever went live with Instant Checkout before OpenAI pivoted in March 2026. The bottleneck was not merchant willingness — Shopify president Harley Finkelstein confirmed the hold-up was on OpenAI's side. The company had not even set up systems to collect and remit state sales taxes, which tells you volumes never reached meaningful scale. ChatGPT now routes purchases through retailer apps (Instacart, Target, Expedia) rather than native product listings. The 4% transaction fee remains. The merchant dashboard does not exist yet.

**Perplexity** took the opposite approach. Zero listing fees. Zero commissions. Zero transaction fees. Merchants keep 100% of revenue from every sale. Perplexity eliminated all advertising in 2026 and is betting its business model entirely on user trust and its Pro subscription. The feed spec requires a Google Shopping CSV delivered via SFTP, with GTINs mandatory — products without UPCs simply do not appear. Real-time pricing accuracy matters heavily because Perplexity cross-references your feed against live site data and suppresses listings with price mismatches. Onboarding takes four to six weeks. For merchants who qualify, the economics are straightforward: it is the only AI shopping channel with no downside cost.

**Google AI Mode and Gemini** are where the most complete infrastructure lives. Google's Universal Commerce Protocol (UCP), launched in January 2026, is an open standard backed by Shopify, Walmart, Target, Etsy, Wayfair, Visa, Mastercard, Stripe, Adyen, Best Buy, Macy's, and The Home Depot. That coalition is not a coincidence. Google has spent 20 years building Merchant Center relationships with every major retailer on earth, and UCP is the protocol that converts those relationships into AI-native commerce. For merchants already running Google Shopping, AI Mode is an additional placement surface that costs nothing incremental. Gemini goes furthest of any platform: it can browse products, compare options, check real-time inventory, and complete purchases via Google Pay without the shopper leaving the interface. It even supports conditional purchases — buy this if the price drops below a threshold. The analytics are the most mature of any platform, reported through Google Ads and Search Console.

**Amazon** is the wildcard. It has the largest product catalog of any retailer on earth, has invested $15 billion in OpenAI, and has explicitly blocked ChatGPT crawlers from accessing its product pages, prices, and reviews. Amazon is building its own AI shopping experience through Rufus, its conversational shopping assistant. The company that controls the most product data in the world has decided not to share it with the AI platforms trying to displace traditional search. That decision will either look prescient or catastrophic in three years.


WHY CHATGPT CHECKOUT FAILED


The three structural problems that blocked ChatGPT Instant Checkout are worth understanding because they apply to every protocol-based approach to AI commerce.

The first is merchant adoption lag. ACP requires merchants to explicitly opt in, integrate their catalogs, and maintain their product data for AI surfaces. Even with Shopify handling the heavy lifting for its merchants, adoption stalled at a handful of large brands. Small and mid-market merchants lack the engineering resources to justify the investment when agent-driven transaction volumes are still negligible.

The second is product data standardization. Merchant product information — pricing, availability, shipping options — needs to be standardized and constantly updated for chatbots to access accurate data. Static catalog feeds go stale. Prices change. Items go out of stock. Any system that relies on pre-ingested catalog data will always be working with information that is at least partially wrong.

The third is fraud and trust safeguards. Merchants and payment firms need safeguards against AI initiating fraudulent or erroneous transactions. Protocol-based systems address this through explicit merchant consent, but that leaves the vast majority of merchants — those who have not opted in — completely out of reach.

Google UCP faces the same structural challenges, but it has two advantages ChatGPT does not. First, it can leverage Merchant Center existing data infrastructure, which already handles real-time inventory and pricing for billions of product listings. Second, its coalition of payment networks and retailers means the protocol has institutional backing that ACP lacks.


WHO WILL WIN


The honest answer is that Google is best positioned to win AI commerce at scale, for reasons that have nothing to do with the quality of its AI.

Google has existing data relationships with every major retailer. It has the payment infrastructure via Google Pay. It has the consumer trust built over two decades of Shopping. It has the most mature analytics of any AI platform. And it has UCP, backed by a coalition that includes every major payment network and most of the largest retailers in the world. The question is not whether Google can win — it is whether antitrust regulators will let it.

ChatGPT has the largest user base but has demonstrated that user base alone does not translate to commerce volume. The checkout pivot is a significant signal: 810 million daily active users and fewer than a dozen merchants live on Instant Checkout is not a commerce business. It is a discovery business with commerce ambitions.

Perplexity has the cleanest merchant economics of any platform and a user base that skews toward high-intent research queries. The zero-fee model is a genuine differentiator. But Perplexity lacks the payment infrastructure, the retailer relationships, and the consumer habit formation that Google has. It is the best early bet for merchants who want to experiment at zero cost.

Amazon will not lose the commerce war. It will fight it on its own terms, with its own AI, on its own platform. The question for brands that sell on Amazon is whether they need to be present in AI platforms that compete with Amazon, or whether Amazon own AI discovery is sufficient. For most consumer brands, the answer is both.


WHAT BRANDS SHOULD DO RIGHT NOW


The practical priority stack for any brand that sells products is straightforward.

Google Merchant Center is table stakes. If you are not already optimized for Google Shopping, start there. AI Mode is an additional placement surface on top of existing Shopping infrastructure, so the work compounds. Add conversational attributes to your product descriptions — answers to common questions, compatible accessories, intended use cases — because these are the fields AI Mode rewards that standard Shopping feeds do not require.

Submit to Perplexity merchant program. It is free, the onboarding is manageable, and the zero-fee economics mean there is no downside. GTINs are mandatory, so ensure your catalog has clean UPC data before applying. Real-time pricing accuracy is critical — Perplexity suppresses listings with price mismatches, so automated feed updates are worth the investment.

Apply to ChatGPT merchant program at chatgpt.com/merchants. The checkout pivot means native transactions are less likely in the near term, but product discovery through ChatGPT is real and growing. Being in the merchant program positions you for whatever commerce infrastructure OpenAI builds next.

Add Schema.org Product markup to every product page. All four platforms use structured data as a signal for product discovery. This is the single highest-leverage technical action for AI commerce visibility across all platforms simultaneously.

For Shopify merchants specifically: enable Shop Pay and ensure your Google channel integration is active. Shopify Agentic Storefronts auto-syndicate your catalog to ChatGPT. The Google channel feeds Merchant Center, which feeds AI Mode and Gemini. The integration work you do once propagates across multiple AI surfaces.

The brands that win AI commerce will not be the ones who waited to see which platform survived. They will be the ones who were already present on all of them when the volume arrived. The volume is arriving now.

Llama, DeepSeek, Qwen, and Mistral now power a third of all AI inference — and every major AEO platform is structurally blind to them. Here is what that costs you, and what to do about it.

Llama, DeepSeek, Qwen, and Mistral now power a third of all AI inference — and every major AEO platform is structurally blind to them. Here is what that costs you, and what to do about it.

Twelve months ago, the combined global usage share of DeepSeek and Qwen was about 1%. By January 2026, it was roughly 15%. That is the fastest adoption curve in AI history. And almost none of it shows up in the dashboards marketing teams use to monitor AI visibility.

When most brands talk about getting recommended by AI, they mean ChatGPT, Gemini, Claude, and Perplexity. Those four are the visible ocean. Below the waterline, a second AI internet has been quietly forming, built on Llama 4, Qwen 3.6, DeepSeek V4, Gemma 4, GLM-5.1, and Mistral Small 4. It powers thousands of consumer apps, internal enterprise copilots, RAG systems behind support portals, and on-device assistants that never call a public API. Whatever you optimized for the closed-model leaderboards is invisible there. And that blind spot is getting bigger, not smaller.


THE TWO AI INTERNETS


There is a useful way to think about the current AI landscape. The closed ecosystem — ChatGPT, Gemini, Claude, Perplexity — is the surface where AI visibility gets discussed because it is the surface where it can be measured. The open ecosystem is the harder half: six open-weight model families, hundreds of forks and fine-tunes, thousands of internal enterprise deployments, and consumer apps in markets most Western brands do not natively monitor. As of mid-2026, that open half accounts for at least a third of all AI inference and is growing faster than the closed half.

Hugging Face now hosts more than 2 million models, 500,000 datasets, and 1 million demo apps. 92.5% of model downloads are for sub-1B parameter models, which means they are being run locally and embedded into products, not called over an API. The mean downloaded model size jumped from 827M parameters in 2023 to 20.8B in 2025, driven by quantization and mixture-of-experts architectures. Open weights are not a research curiosity anymore. They are the substrate of an entire production layer of AI.


WHY OPEN MODELS SEE YOUR BRAND DIFFERENTLY


Both closed and open ecosystems start from Common Crawl. 64% of the 47 LLMs Mozilla analyzed used at least one filtered version of it. What diverges is what each model does with that raw data. Closed labs run private quality classifiers and layer in curated proprietary data. Open models like Llama 4, Qwen, and DeepSeek lean on public filtered datasets such as RedPajama-V2, FineWeb, and Dolma. These public pipelines are aggressive about deduplication and quality scoring, which is good for model performance but punishing for brands whose web presence is thin or lives mostly on their own domain.

Common Crawl itself is not a representative sample of the web. Its crawler prioritizes domains that are heavily linked to, so Facebook, Google, YouTube, and Wikipedia dominate the graph and long-tail industry sources get sparse coverage. When public filtering pipelines layer on top of that bias, brands that are well known to ChatGPT can vanish from open-weight model recall entirely.

Then there is quantization. The 4-bit and 8-bit quantized versions of Llama 4 and Qwen that ship into production apps trade fidelity for speed. The first capability to degrade is long-tail entity recall — which is exactly where most B2B brands live. Your brand might be in the full-precision weights and silently absent from the version a developer actually deployed.


THE SELF-HOSTED ENTERPRISE BLIND SPOT


Gartner reported a 340% jump in enterprise private LLM development through 2025. 65% of Fortune 500 companies have deployed LLM-based engagement tools. 44% of organizations cite data privacy as the top barrier to using public AI — the exact wedge that pushes them toward self-hosted Llama, Qwen, and Mistral deployments inside the corporate firewall.

From an AEO perspective, this is a category change, not an incremental one. When a Fortune 500 buyer asks their internal procurement copilot which vendors to shortlist, that copilot is often a fine-tuned Llama or Qwen running inside the corporate firewall. There is no API to monitor. No prompt logs to scrape. No public ranking to track. If your brand was not in the base model and is not in the RAG corpus the company assembled, you are not on the shortlist. And the buyer has no idea you exist.


THE DARIO PROBLEM


There is a political dimension to this story that the AEO industry has largely ignored. Anthropic CEO Dario Amodei has been among the most vocal advocates for restricting open-weight models. He told the US Senate that open source AI is on a dangerous path. In June 2026, he published a lengthy essay arguing that governments should have the power to block AI models from deployment if they present unacceptable risks. OpenAI and Anthropic have since aligned in Washington on warning policymakers about the risks of powerful Chinese open-weight models specifically.

https://youtu.be/6Yl3EvPqPgY?si=_Ae0VT6mphFNkvqF

_blank

noopener noreferrer

text-amber-700 underline hover:text-amber-900

Nathan Lambert at Interconnects AI put it plainly in July 2026: the action Anthropic is effectively asking for is the wholesale banning of pretty much all Chinese open-weight models in the US. That would demolish the open model economy that is emerging with inference companies, fine-tuning companies, and new products built on top of them. It would also be futile. If the models are not banned in China as well, it is very easy for a bad actor to still use a banned open-weight model, which negates the safety argument entirely.

The critics of this agenda include David Sacks, the Trump administration AI adviser, who has said the scrutiny could amount to regulatory capture: rules intended to improve AI safety could instead entrench the largest companies by making it harder for competitors to release models. That framing is correct. The closed labs stand to benefit enormously from greater scrutiny of open-weight models. That conflict of interest should be front of mind whenever Anthropic or OpenAI publish safety arguments about open-source AI.

My view is simple: open always wins. It won in operating systems. It won in databases. It won in cloud infrastructure. The idea that AI will be the exception — that a handful of closed labs will permanently control access to intelligence — is not a safety argument. It is a business model dressed up as one.


RAG CHANGES THE VISIBILITY MATH


Most production deployments of open-weight models are not bare. They sit behind a retrieval layer. Documents get embedded, queries hit a vector database, and the LLM synthesizes an answer from retrieved chunks plus its parametric knowledge. The closed-model conversation about training data presence matters less here. What matters is whether your content was ingested into the RAG index.

That sounds like an opportunity, and it is — partially. If a developer wires your docs into their internal copilot, you get cited every time. But the practical pattern is that companies index their own internal docs, the public docs of incumbents they already trust, and a curated knowledge base. New entrants are absent from both the model weights and the RAG corpus. The asymmetry compounds.

The brands that win this layer ship their content in formats that get pulled into RAG indexes by default. That means clean Markdown documentation, public API references with structured schemas, llms.txt files, and content that is easy to chunk and embed. If your top-of-funnel page is a JavaScript-rendered marketing site with no direct factual claims, you are invisible to retrieval too.


WHAT TO ACTUALLY DO ABOUT IT


The practical response is not complicated, but it requires treating open-weight model visibility as a separate workstream from closed-model AEO, because the optimization levers are different.

Start by testing your visibility on at least one open model directly. Run Qwen 3.6 or Llama 4 against your category prompts via Hugging Face Inference, Together AI, Groq, or Fireworks. The cost is trivial and the signal is real. If your brand surfaces on ChatGPT but not on Llama 4, you have a training-data gap, not a hallucination problem.

Then audit your presence in the public sources that public filtering pipelines preserve. Wikipedia is not optional. Crunchbase, GitHub READMEs that describe what you do, structured Schema.org markup, and at least one canonical entry on each of the major review aggregators in your category are the sources RedPajama, FineWeb, and Dolma are biased toward keeping.

Publish for retrieval. Maintain a comprehensive plain-Markdown documentation site, expose llms.txt and llms-full.txt, and structure your highest-value pages so they survive chunking. If you only have one set of marketing pages and they are JavaScript-rendered, you are double-blind on the open layer.

If you sell internationally, earn citations in non-Western sources. A single Zhihu post or Baidu Baike entry does for Qwen what a TechCrunch piece does for ChatGPT. The same logic applies to NAVER for Korean models.

And accept that the self-hosted enterprise layer will stay opaque. You cannot monitor what runs inside a corporate firewall. What you can do is track inbound traffic for AI user agents you do see (PerplexityBot, GPTBot, ClaudeBot), monitor support tickets and sales calls for mentions of internal AI tools, and instrument your demo flow for unusual referral patterns. Self-hosted copilots cannot be queried, but their downstream behavior leaves traces.

The closed-model leaderboard is the easy half of AEO. The brands that recognize the open layer early and treat training-data presence as a portfolio play across both ecosystems will have a structural advantage over those that keep optimizing for the dashboards and wonder why their pipeline numbers do not move.

The 2026 AI Visibility Index is the largest public dataset on AI citation patterns to date. ChatGPT cites 15 sources per response on average. Gemini cites 3. Only 36 global brands maintained top-100 visibility across all four platforms studied.

The 2026 AI Visibility Index is the largest public dataset on AI citation patterns to date. ChatGPT cites 15 sources per response on average. Gemini cites 3. Only 36 global brands maintained top-100 visibility across all four platforms studied.

<a href="https://www.semrush.com/news/463141-semrush-releases-expanded-2026-ai-visibility-index-analyzing-126-million-ai-search-prompts/" target="_blank" rel="noopener noreferrer" class="text-amber-700 underline hover:text-amber-900">Semrush's 2026 AI Visibility Index</a> — the largest public study of AI citation patterns to date — analysed 126 million AI search prompts across 16 industries and four major AI platforms: ChatGPT, Gemini, Perplexity, and Claude. The findings challenge several assumptions that have become conventional wisdom in the AEO space.

The most striking data point is the citation gap between platforms. ChatGPT cites an average of 15 sources per response. Gemini cites an average of 3. This is not a minor difference — it means a brand that appears consistently in ChatGPT answers may be almost invisible in Gemini, and vice versa. Optimising for 'AI search' as a monolithic channel is a category error. Each platform has a distinct citation logic.


THE INTEGRATION PREMIUM


The study's most actionable finding is the integration premium. Organisations with integrated SEO and AI visibility strategies reported increased traffic or leads from AI platforms at a rate of 81%. Organisations managing SEO and AI visibility as separate workstreams reported the same outcome at a rate of 36%. The gap — 45 percentage points — is the largest performance differential in the study.

Only 36 global brands maintained top-100 visibility across all four platforms studied. These brands share three characteristics: high domain authority, consistent third-party citation presence, and structured content that answers specific buyer questions directly.


THE MEASUREMENT GAP


45% of marketing leaders surveyed cannot accurately measure their brand's visibility in AI-generated answers. This is the most significant operational gap in the industry right now. Brands that cannot measure AI visibility cannot manage it — and the Semrush data suggests that the gap between measured and unmeasured brands is widening as AI Mode usage accelerates.

The AI traffic growth numbers are striking. Adobe data cited in the report shows AI-driven traffic to US retail sites surged 1,324% between October 2024 and May 2026. In the travel sector, the figure is 2,215%. These are not incremental shifts — they represent a fundamental reallocation of discovery traffic from traditional search to AI-generated answers.

One year after its debut, Google AI Mode has crossed a milestone that reframes the entire AEO conversation. The average AI Mode query is now triple the length of a traditional search — and planning queries are growing 80% faster than overall AI Mode usage.

One year after its debut, Google AI Mode has crossed a milestone that reframes the entire AEO conversation. The average AI Mode query is now triple the length of a traditional search — and planning queries are growing 80% faster than overall AI Mode usage.

Google announced at <a href="https://blog.google/products-and-platforms/products/search/search-io-2026/" target="_blank" rel="noopener noreferrer" class="text-amber-700 underline hover:text-amber-900">Google I/O 2026</a> that AI Mode has surpassed 1 billion monthly users globally — one year after its debut. AI Mode queries have more than doubled every quarter since launch, making it the fastest-growing product in Google Search history.

The scale of the shift is visible in how people are searching. The average AI Mode query is now triple the length of a traditional search query. More than one in six searches in the US now use voice or images. Image searches are growing over 40% month-over-month. Planning queries — 'help me plan a trip to Tokyo', 'what should I consider when switching CRM providers' — are growing 80% faster than overall AI Mode queries.


WHAT THIS MEANS FOR AEO


The 1 billion user milestone is not just a product announcement. It is a structural shift in how brands get discovered. When search queries triple in length and shift toward planning and comparison, the brands that get cited are those that have invested in comprehensive, authoritative, conversational content — not those optimised for short-tail keywords.

The 80% growth in planning queries is particularly significant for B2B brands. Planning-stage queries — 'what should I look for in an AEO platform', 'how do I evaluate AI visibility tools' — are where vendor selection begins. Brands that appear in AI Mode answers at the planning stage are building consideration before the buyer has even formed a shortlist.

Google's own data suggests that AI Mode is not cannibalising traditional search — it is expanding the scope of what search can answer. For AEO practitioners, the implication is clear: the question is no longer whether to optimise for AI answers. It is how quickly you can build the content depth and third-party authority that AI Mode requires to cite your brand.

A niche AI visibility tracker in the wellness space ran 90 queries across ChatGPT, Perplexity, and Claude. Not one open-ended question returned an unprompted brand recommendation. The sample is small — but the pattern matches broader research.

A niche AI visibility tracker in the wellness space ran 90 queries across ChatGPT, Perplexity, and Claude. Not one open-ended question returned an unprompted brand recommendation. The sample is small — but the pattern matches broader research.

A new micro-study from <a href="https://breadchaser.ai/blog/ai-visibility-report-q3-2026" target="_blank" rel="noopener noreferrer" class="text-amber-700 underline hover:text-amber-900">Breadchaser.ai</a>, an AI visibility tracking tool focused on the health, fitness, and wellness sector, tested three businesses across 90 AI-generated answers — 10 buyer-intent prompts per business, run across ChatGPT, Perplexity, and Claude during the week of July 7–13, 2026.

The finding is stark: across all 90 answers, there were zero unbranded brand recommendations. Every one of the 17 brand mentions that did appear came from queries that already contained the brand name. When questions were open-ended — "best [category]", "which tool should I use for X" — the AI engines named someone else every time.


WHAT THE AI CITED INSTEAD


In the agency category, Reddit was the most-retrieved domain (32 retrievals in a single 10-prompt run). YouTube appeared 20 times in both the agency and wellness product runs. LinkedIn appeared 23 times in the agency run. In the sauna/wellness product category, the market leader's own comparison listicles were retrieved 33 times — the single most effective AI visibility asset found in the entire study.

The study also surfaced what founder Nick Montes calls a "trap door" finding: asking an AI about a specific brand by name — "[brand] reviews" — caused one engine to recommend a competitor, because the brand had a thin review and listing footprint. Being named in a query is not enough to guarantee a positive citation.


THE CAVEATS


The sample is deliberately narrow: three businesses, one industry vertical, 90 answers. Gemini was excluded due to API quota constraints. The study was conducted by a vendor whose product is designed to solve the exact problem the study identifies. These are important limitations.

But the directional finding is consistent with larger-scale research. The <a href="https://tryorbt.com/publications/the-recommendation-economy" target="_blank" rel="noopener noreferrer" class="text-amber-700 underline hover:text-amber-900">ORBT Recommendation Economy report</a> (July 2026), which reviewed 30+ studies covering hundreds of millions of citations, found that technical GEO tactics have "weak, null, or negative" relationships with AI visibility. What does work — across both the Breadchaser micro-study and the broader literature — is third-party citation presence, review footprint, and structured content on high-authority domains.

The Breadchaser study is not definitive. But it asks the right question: if your brand cannot be recommended without being named first, is it actually visible?

A survey of 1,038 B2B software buyers finds that AI has compressed the shortlist stage but intensified scrutiny at evaluation — and that brands missing from AI answers are eliminated before buyers ever reach out.

A survey of 1,038 B2B software buyers finds that AI has compressed the shortlist stage but intensified scrutiny at evaluation — and that brands missing from AI answers are eliminated before buyers ever reach out.

https://www.prnewswire.com/news-releases/ai-is-making-software-easier-to-find-and-harder-to-buy-according-to-new-g2-research-302831346.html

_blank

noopener noreferrer

text-amber-700 underline hover:text-amber-900

82% of B2B software buyers sourced software recommendations from an AI chatbot in the last 24 months. But once a shortlist forms, the friction intensifies. Evaluation now accounts for 40% of the total buying timeline, surpassing research (down to 36%) as the longest stage. IT security review is the single biggest source of post-selection delay, cited by 39% of buyers and rising to 50% among enterprise buyers. Finance involvement in software decisions jumped from 31% to 46% in a single year, and 49% of buyers say their CFO vetoed an already-approved software purchase in the last 12 months.


THE AEO IMPLICATION: SHORTLIST OR INVISIBLE


The finding with the most direct AEO relevance is the shortlist dynamic. Among buyers who used AI chatbots, 50% said the greatest impact was narrowing and comparing options. Review sites (38%) have now overtaken AI chatbots (37%) as the top source influencing which vendors make the initial shortlist — the first time review platforms have led that ranking. Brands that do not appear in AI answers and review sites are eliminated before a buyer ever reaches out.

G2 notes that it has become the most-cited B2B software source across AI-first channels, a position that reflects the broader pattern identified in large-scale citation research: AI engines weight third-party, high-credibility review sources heavily when answering vendor evaluation queries. For B2B brands, the practical implication is that G2 review presence is now an AEO asset, not just a sales tool.


PRICING AND CONTRACTS UNDER PRESSURE


AI has also disrupted B2B pricing expectations. Preference for outcome-based pricing more than doubled in a single year, from 11% to 23%. 80% of organizations now provide developers or technical teams with a dedicated LLM usage budget, creating a new line item that procurement teams are scrutinising alongside traditional seat licenses. 70% of buyers say the pace of technology innovation is pushing them toward shorter contracts.

On the question of AI agents in the buying process: 61% of buyers currently use or plan to use AI agents as part of software evaluation, primarily for total cost of ownership analysis, shortlist building, and vendor research. But only 9% are comfortable letting agents execute purchases within approved guardrails, and just 2% would allow purchases without pre-approval. The agentic buying era is approaching, but human sign-off remains firmly in place.

New data from Funnel, covering 5,000+ companies and 11% of global digital ad spend, finds paid search retaining its top spot in the marketing mix despite the rise of answer engine optimization.

New data from Funnel, covering 5,000+ companies and 11% of global digital ad spend, finds paid search retaining its top spot in the marketing mix despite the rise of answer engine optimization.

https://www.thedrum.com/news/as-answer-engine-optimization-rises-traditional-paid-search-is-still-growing-for-now

_blank

noopener noreferrer

text-amber-700 underline hover:text-amber-900

The growth is concentrated, as it has been for years, in two platforms. Google Ads posted year-on-year increases of 112% in EMEA to $110 million and 41% in the US to $99 million. Facebook Ads grew 148% in EMEA to $62 million and 211% in the US to $84 million. Those figures align with Alphabet's Q2 2026 earnings, published July 22, which showed Google Search and Other revenue up 17% year over year to $63.27 billion — a rate that eased from 19% in Q1 but remains well above the pace most analysts expected given the scale of AI search adoption.


SOCIAL GROWS EVEN FASTER — BUT TRAILS SEARCH


Social media spend grew faster than paid search in the same period, at 150% from $26 million to $65 million. In EMEA, social has now claimed the second spot in digital ad spend rankings, displacing display advertising into third place. TikTok's EMEA spend rose from $2 million in January 2024 to $12 million by April 2026, placing it in the top five platforms in the region for the first time. The US is described as on the same trajectory, a few quarters behind.

APAC growth was the most dramatic in the dataset, at 400% across the period. November remains the single largest month for digital ad spend globally, with EMEA Black Friday receipts doubling from $125 million in 2024 to $270 million in 2025.


WHAT THIS MEANS FOR AEO INVESTMENT


The Funnel data does not challenge the case for AEO investment — it contextualises it. Paid search is growing because it is measurable, attributable, and fast. AEO is growing because it is building the brand authority that determines which brands get recommended before a buyer ever runs a paid search. The two disciplines are not substitutes. Brands that treat them as competing budget lines are misreading the dynamic.

The more significant signal in the data is what it implies about the transition timeline. If paid search spend is still accelerating at 122% despite widespread awareness of AI search's rise, the brands that have already begun building AEO infrastructure are not early adopters — they are simply ahead of a curve that is still forming. The window for first-mover advantage in AI citation is open, but the Funnel data suggests it will not stay open indefinitely.

The foundation is the same. The crawlability, the authority signals, the structured data — all of it still matters. But AEO adds a layer that SEO was never built for: the claims layer. That is where the real work begins.

The foundation is the same. The crawlability, the authority signals, the structured data — all of it still matters. But AEO adds a layer that SEO was never built for: the claims layer. That is where the real work begins.

The framing that AEO and SEO are opposites is wrong, and it is causing real strategic harm. Brands that accept the binary either dismiss AEO as a rebrand of what they already do, or they abandon their SEO program in favor of something new and unproven. Both responses are mistakes. The accurate picture is more useful and more demanding: AEO is built on the same foundation as SEO, but it adds a layer that SEO was never designed to address. Understanding exactly where the overlap ends and the new territory begins is the most important strategic question in search marketing right now.

<h2>What Carries Over: The Shared Foundation</h2>

Start with what is genuinely shared, because it is substantial. Technical SEO is not optional for AEO. A page that cannot be crawled cannot be cited. A site with poor Core Web Vitals loads slowly for AI crawlers too. Canonical tags, robots.txt, sitemap structure, and clean URL architecture are prerequisites for AI visibility in exactly the same way they are prerequisites for Google indexing. The brands that neglected technical SEO will find their AEO programs built on sand.

Structured data is the most direct bridge between the two disciplines. Schema markup, which SEO practitioners have used for years to help Google parse page content, is now one of the most reliable signals for AI citation. JSON-LD implementations of Article, FAQPage, HowTo, and Organization schema give AI engines a machine-readable summary of what a page claims to be about. Sites implementing FAQ schema are seeing significantly higher AI citation rates than those relying on unstructured content alone. [1] The investment made in structured data for SEO is not wasted in AEO. It is amplified.

E-E-A-T, Google's framework for evaluating Experience, Expertise, Authoritativeness, and Trustworthiness, was developed for organic search but its logic maps almost perfectly onto AI citation behavior. AI engines weight content from named, credentialed authors more heavily. They favor sources that cite primary research. They deprioritize content that makes claims without evidence. The E-E-A-T investments that improved Google rankings -- author bios, original research, cited statistics, editorial standards -- are the same investments that improve AI citation rates. [2]

Page speed, mobile optimization, and accessibility also carry over. AI crawlers are not more patient than Googlebot. And semantic HTML, which accessibility advocates have championed for years, turns out to be critical for AI agents that read the accessibility tree rather than rendered pixels. A styled div that looks like a heading is invisible to an AI agent. A properly tagged h2 is not.

<h2>The Four-Layer AEO Model</h2>

The clearest way to understand the relationship between SEO and AEO is through what we call the Four-Layer AEO Model. The bottom two layers are shared with SEO and represent the foundation every brand must get right before AEO work can compound. The top two layers are genuinely new territory: the claims and entity layer, and the answer architecture and measurement layer. Brands that stop at Layer 2 will rank in Google. Brands that build all four layers will be cited in AI answers.

<img src="/manus-storage/four_layer_model_generated_3fffca23.webp" alt="The Four-Layer AEO Model" class="w-full rounded-lg my-6" />

<h2>The Overlap: What Carries Over vs. What Is New</h2>

OVERLAP_TABLE

<h2>Where AEO Ventures Into New Territory</h2>

The shared foundation is real, but it only gets a brand to the starting line. The genuinely new territory in AEO is what can be called the claims layer: the set of signals that determine whether an AI engine trusts a brand's statements enough to repeat them as facts.

In SEO, the authority question is about links. Who points to you? How many? From what domains? In AEO, the authority question is about claims. What does your brand assert? Are those assertions corroborated by independent sources? Do the third-party pages that mention your brand describe you accurately and specifically? A brand can have a strong backlink profile and still be invisible in AI answers if the text of those links does not contain the right claims about the right capabilities.

This is the inversion that most SEO practitioners miss. In traditional SEO, you optimize your own pages to rank. In AEO, the most important optimization work happens on pages you do not own. Whitespark's research on local AI citations found that 60% of AI Overview citations point to third-party publishers, not the brand's own domain. [3] The Breadchaser study of 90 AI answers found zero unbranded mentions -- every citation was tied to a named entity with a verified identity. [4] The implication is that your brand's AI visibility is largely determined by how third-party sources describe you, not by how you describe yourself.

<h2>A Real-World Example: The Brand That Ranks First but Gets Cited Never</h2>

Consider a mid-market CRM company. It has invested heavily in SEO for five years. It ranks position one for "best CRM software for teams." Its technical SEO is clean, its backlink profile is strong, and its blog publishes twice a week. By every traditional metric, it is winning.

Ask ChatGPT, Perplexity, or Gemini the same question. The brand does not appear. Salesforce, HubSpot, and Notion are cited instead -- not because they rank higher in Google, but because they have built the claims layer. Their capabilities are described consistently across hundreds of third-party review sites, analyst reports, and editorial pieces. Their entities are clean and well-defined in every knowledge graph. Their content is structured to answer specific questions directly, not to cover broad topics comprehensively. The CRM company has the foundation. It has not built the structure on top of it.

<img src="/manus-storage/brand_example_generated_9887f8a1.webp" alt="Ranking First Does Not Mean Being the Answer" class="w-full rounded-lg my-6" />

The entity layer is the second genuinely new territory. SEO has always cared about keywords. AEO cares about entities: named, verifiable things that AI engines can anchor to a knowledge graph. A brand that is a well-defined entity -- with a consistent name, a clear description, a verified Wikipedia or Wikidata entry, and consistent NAP data across the web -- is far more likely to be cited than a brand that exists only as a keyword on its own pages. Entity clarity is not a new concept in SEO, but it has never been as consequential as it is in AI search.

The third new territory is the answer architecture of content itself. SEO rewards comprehensive coverage. AEO rewards extractability. A 3,000-word pillar page that covers a topic broadly may rank well in Google and be completely ignored by every AI engine. A 600-word piece that answers a precise question in the first sentence, cites a named study, and structures each section as a direct response to a specific query is far more likely to be cited -- even if it never appears on the first page of Google. The Princeton GEO study found that keyword stuffing decreased AI visibility by 10 percent, while adding statistics and citations increased it significantly. [5] The content that wins in AI search is not the content that covers the most ground. It is the content that is most confidently right about a specific thing.

<h2>The Metrics Diverge</h2>

The clearest sign that AEO is a distinct discipline is that its success metrics do not exist in any SEO dashboard. SEO teams report on rankings, organic traffic, and click-through rates. None of these metrics have a direct equivalent in AI search. When ChatGPT recommends your brand, there is no ranking to track, no click to count, and no impression to measure in Google Search Console.

The relevant AEO metrics are share of voice across AI platforms, citation frequency by query category, sentiment accuracy in AI-generated descriptions, and domain territory -- the semantic space in which your brand is mentioned and recommended. These require a different measurement stack entirely. Platforms like Profound, Otterly AI, and Peec AI were built specifically to track these signals because the existing SEO toolset cannot.

<img src="/manus-storage/metrics_diverge_generated_74797421.webp" alt="Where SEO Metrics End and AEO Metrics Begin" class="w-full rounded-lg my-6" />

<h2>The Practical Implication</h2>

The right mental model is not SEO versus AEO. It is SEO as the foundation and AEO as the structure built on top of it. A brand that has neglected technical SEO, structured data, and E-E-A-T will struggle in AEO. A brand that has invested in those things but has not addressed the claims layer, the entity layer, and the answer architecture of its content will rank well in Google and be invisible in AI answers.

The brands that will win in AI search are the ones that refuse the false binary. They maintain and improve their SEO program because the foundation still matters. And they build the AEO layer on top of it: managing their entity presence, engineering their content for extractability, earning the third-party mentions that carry the right claims, and measuring success with metrics that SEO tools were never designed to capture.

SEO earns you a place in the list. AEO earns you the right to be the answer. The foundation is shared. The destination is not.

<h3>References</h3>

[4] Breadchaser, Small Study, Sharp Finding: Zero Unbranded Brand Mentions Across 90 AI Answers, 2026.

[5] Princeton NLP Group, Generative Engine Optimization (GEO) study, 2024.

A new meta-analysis of 45 GEO studies finds no consistent improvement in AI discoverability. The study is largely correct. The problem is that it is evaluating the wrong thing entirely.

A new meta-analysis of 45 GEO studies finds no consistent improvement in AI discoverability. The study is largely correct. The problem is that it is evaluating the wrong thing entirely.

https://arxiv.org/abs/2311.09735

_blank

noopener noreferrer

text-amber-700 underline hover:text-amber-900

40% visibility increase

https://tryorbt.com/publications/the-recommendation-economy

_blank

noopener noreferrer

text-amber-700 underline hover:text-amber-900

40% visibility increase

The response from much of the AEO industry has been defensive. It should not be. The study is largely correct. The problem is that it is evaluating the wrong thing entirely.


WHAT THE STUDY ACTUALLY FOUND


The 45 studies reviewed share a common design: take a piece of content, apply a GEO intervention — rewrite a paragraph, add schema markup, restructure a heading — then measure whether AI engines cite that content more frequently. The meta-analysis finds these interventions produce inconsistent results. Sometimes a modest improvement, sometimes no change, occasionally a decline.

This is an accurate finding. Anyone who has spent serious time working on AI visibility already knows that paragraph rewrites and keyword insertions do not reliably move citation rates. The problem is not the conclusion. It is the premise. The studies are measuring the wrong lever entirely.


IT IS NOT A CONTENT PROBLEM. IT IS A CLAIMS PROBLEM.


AI citation is not primarily a content formatting problem. It is a <em>claim architecture</em> problem — and beneath that, a <em>domain ownership</em> problem.

When an AI engine decides which brand to cite in response to a purchase-relevant question, it is not scanning for well-structured paragraphs. It is looking for sentences that are <em>specific</em>, <em>defensible</em>, and <em>citable</em> — the kind of claim it can extract, repeat, and present as a recommendation with confidence. A specific percentage backed by a clinical study. A named certification no competitor holds. A concrete outcome tied to a defined time frame. These are the sentences AI engines lift and repeat. Broad quality assertions and mission statements are the sentences they skip.

The meta-analysis reviewed studies that tested whether restructuring existing content around better formatting improves citation rates. What it did not test — because it is nearly impossible to isolate in a controlled academic setting — is whether <em>publishing fundamentally different claims</em> changes citation behaviour. That is the actual intervention that matters.


THE DOMAIN TERRITORY QUESTION


Before a brand can publish citable claims, it needs to answer a more fundamental question: <strong>what territory do we want to own in AI search?</strong>

This is not a content strategy question. It is a brand strategy question. And most brands have not answered it.

Consider a well-known supplement brand competing in a crowded consumer health category. Its market position is strong — second in web footprint in its category, significant retail distribution, a parent company with over a century of scientific credibility. By every traditional metric, it should be earning AI citations at a rate commensurate with its market share.

Instead, it earns a fraction of what its smaller, digitally-native competitors earn. The AI gap — the difference between its market position and its AI share of voice — is the largest in its category.

The reason is not technical. Its website is fast, well-structured, and properly marked up. The reason is that its <em>claims</em> are the wrong type. Its content is full of broad, unanchored assertions about product quality and brand heritage — the kind of language that reads well to a human and means nothing to an AI engine looking for something citable.

Meanwhile, its most-cited competitor earns a disproportionate share of all category AI citations from a single sentence. That sentence combines a specific percentage, a named user group, a defined outcome, and a time horizon. It is not better written than anything the first brand publishes. It is <em>structurally different</em> — specific, defensible, and extractable in a way that general brand language never will be.

The first brand's biggest differentiator — its parent company's scientific heritage, its regulatory approval, its clinical backing — exists as a vague brand impression, not as a citable claim. No amount of paragraph rewriting fixes that. The fix requires stepping back from content tactics entirely and asking: what is the specific, defensible territory this brand should occupy in AI-generated answers? What are the two or three sentences that, if published and amplified, would make AI engines reliably associate this brand with that territory?


WHAT WE HAVE FOUND WITH CLIENTS


The brands that improve AI share of voice most significantly are not the ones that optimise their existing content. They are the ones that do two things first.

<strong>They audit their claim architecture.</strong> Not their website structure — their <em>claims</em>. What specific, defensible statements does this brand have the right to make? What data exists — clinical, commercial, operational — that has never been expressed in a form an AI engine can extract? What is the brand entitled to say that no competitor can say?

<strong>They identify the domain they want to own.</strong> Not a broad category, but a specific, defensible territory. "The only platform in this category with a published third-party security audit and SOC 2 Type II certification." "The only supplement in this category formulated by a global pharmaceutical company with regulatory approval." That territory becomes the brief for every piece of content that follows.

One client — a B2B software company in a crowded enterprise category — had a strong product, a credible customer base, and a well-maintained website. Its AI share of voice was a fraction of its market share. The gap was not technical. Its content was full of feature descriptions and general value propositions. It had no citable claims.

We identified the domain it could defensibly own, engineered a small number of specific, high-weight sentences around that territory, and published them across the website and in earned media placements. Within 90 days, its AI share of voice had moved significantly. The content did not get better written. The claims got more specific, more defensible, and more aligned with what AI engines actually extract.


WHAT THE STUDY GETS RIGHT — AND WHAT IT MISSES


The meta-analysis is correct that GEO as it is currently practised — paragraph rewrites, schema additions, keyword insertions — does not consistently improve AI visibility. That is a fair and important finding. There is a great deal of tactical noise in this space, and the study is right to challenge the easy claims.

What it misses is the distinction between <em>content optimisation</em> and <em>claim architecture</em>. The studies it reviews are almost entirely testing the former. The latter — identifying the domain a brand wants to own, auditing what citable claims exist and which are missing, then publishing specific, defensible sentences that AI engines can extract — is not something that can be tested in a controlled academic study. It requires brand strategy, competitive analysis, and editorial rigour working together.

Notably, the study's own recommendation is to "optimise content quality and authority." That is exactly what serious AEO looks like. The authors have inadvertently validated the approach while dismissing the tactics. Those are not the same thing.


THE HARDER TRUTH


The deepest implication of both the study and the client work is one that most brands are not ready to hear: AI visibility is largely a function of real-world brand authority, not content optimisation.

A brand that has published original research, earned third-party citations from authoritative sources, built a genuine entity presence in knowledge graphs, and accumulated a track record of specific, verifiable claims will earn AI citations. A brand that has done none of those things but has well-structured schema markup will not.

This is not a technical problem. It is a brand-building problem — one that happens to manifest in AI search results. The brands winning AI citations are not winning because their developers added FAQ schema. They are winning because they have spent years publishing specific, sourced, citable content that answers the questions their buyers ask.

The study asked whether GEO techniques work. The more important question is: what kind of brand do you need to be for AI engines to want to cite you? And what specific claims, published in what specific form, will get you there fastest?

Those are the questions worth answering.

Consumer and SMB AEO is about being cited. Enterprise AEO is about being trusted by a buying committee that never visits your website. The tactics are different. Here is the framework.

Consumer and SMB AEO is about being cited. Enterprise AEO is about being trusted by a buying committee that never visits your website. The tactics are different. Here is the framework.

Most of the AEO literature published in the past two years has been written with a single user in mind: a person who types a question into ChatGPT or Perplexity and receives an answer. That user might be a consumer choosing a restaurant, a small business owner evaluating software, or a professional researching a topic. The AEO playbook built for that user — optimize for extractability, earn third-party citations, build entity signals, structure content for direct answers — is sound. But it is incomplete.

When the buyer is not an individual but a committee, when the purchase cycle runs six to eighteen months rather than six minutes, and when the AI surfaces being consulted are not just public search engines but internal enterprise tools, the strategy changes in ways that most AEO frameworks have not yet accounted for.

This article is an attempt to name those differences precisely, and to describe what an enterprise-grade AEO strategy looks like in practice.


THE FUNDAMENTAL DIFFERENCE: WHO IS ASKING THE QUESTION


In consumer and SMB contexts, AEO is a one-to-one discipline. A single user asks a question; a single AI answers it; a single citation either appears or does not. The buying decision that follows is largely individual. The AEO goal is to be present in that single moment of AI-mediated discovery.

Enterprise buying does not work this way. A typical enterprise software purchase involves six to ten stakeholders, each asking different questions of different AI systems at different stages of a multi-month evaluation. The CFO asks ChatGPT about total cost of ownership. The CISO asks Perplexity about security certifications. The procurement team uses ChatGPT Business Search to summarise vendor comparison documents saved in their SharePoint. The end users ask colleagues in Slack, who have their own AI assistants surfacing recommendations from their browsing history and saved documents.

Each of these is a separate AI citation surface. Each requires a different kind of content to perform well. And unlike the consumer context, where a single citation in a popular AI engine can drive thousands of visits, the enterprise context is about depth of presence across a small number of high-stakes interactions — the moments when a specific stakeholder, at a specific stage of an evaluation, asks a specific question.


THE FIVE SURFACES THAT MATTER IN ENTERPRISE AEO


Enterprise AEO requires visibility across five distinct surfaces, each with its own retrieval logic and content requirements.

**Public AI answer engines** (ChatGPT Search, Perplexity, Google AI Mode, Bing Copilot) remain important, but their role in enterprise buying is different from consumer buying. Enterprise buyers use public AI engines primarily in the early stages of a purchase cycle — category education, vendor landscape mapping, and initial shortlisting. The content that performs well here is definitional and comparative: clear explanations of what the product does, how it differs from alternatives, and what category of problem it solves. Jargon-heavy product marketing copy performs poorly; direct, structured explanations of capability perform well.

**Internal enterprise AI tools** — ChatGPT Business Search, Microsoft 365 Copilot, Google Workspace Gemini, and Salesforce Einstein — are the surfaces that most AEO strategies have not yet addressed. These tools retrieve content from documents that enterprise employees have saved, shared, or connected to their workspace. A brand that is well-represented in the documents enterprise buyers collect and share — analyst reports, comparison guides, case studies, RFP templates — will surface in these internal AI answers. A brand that is not in those documents will not surface, regardless of its public-web citation rate.

**Analyst and research platforms** (Gartner, Forrester, IDC, G2, Capterra) are a critical intermediary surface. Enterprise buyers treat analyst reports as authoritative sources, and they save, share, and annotate them extensively. When an internal AI tool retrieves content from a saved Gartner report, the brands mentioned in that report gain citation presence inside the enterprise's AI ecosystem. Earning a named mention in a Gartner Magic Quadrant or a Forrester Wave is not just a sales tool — it is an AEO asset that propagates through every enterprise that subscribes to that research.

**Peer review and community platforms** (G2, TrustRadius, Reddit, LinkedIn) are increasingly cited by public AI engines when answering vendor evaluation questions. Perplexity, in particular, surfaces G2 reviews and Reddit threads prominently in response to queries like 'what do users think of [vendor]' or 'is [vendor] worth it for enterprise'. The content that appears in these citations is user-generated and largely outside the brand's direct control — but the brand can influence it by ensuring that satisfied customers are prompted to leave structured, detailed reviews that include the specific use cases and outcomes that enterprise buyers search for.

**The brand's own content ecosystem** — website, blog, documentation, case studies, and help content — remains the foundation. But in enterprise AEO, the content hierarchy is different from consumer AEO. The pages that drive enterprise AI citations are not the homepage or the product overview page. They are the detailed technical documentation, the security and compliance pages, the integration directories, the ROI calculators, and the case studies that describe specific enterprise deployments with named outcomes. These are the pages that enterprise buyers save, share with colleagues, and ask their AI tools to summarise.


THE CONTENT STRATEGY DIFFERENCE


Consumer AEO content is optimised for a single question asked by a single user. Enterprise AEO content must be optimised for a matrix of questions asked by different stakeholders at different stages of a buying cycle.

The most useful framework for enterprise AEO content is the buying committee map. For each product or solution, identify the six to ten roles that typically participate in an enterprise purchase decision: the economic buyer, the technical evaluator, the security reviewer, the legal and compliance reviewer, the end user champion, and the executive sponsor. For each role, identify the three to five questions they are most likely to ask an AI engine during their evaluation. Then audit whether the brand's existing content provides a direct, extractable answer to each of those questions.

In most cases, the audit reveals significant gaps. Security reviewers ask questions about SOC 2 certification, data residency, and penetration testing — and most vendor websites bury this information in a compliance PDF that AI crawlers cannot access. Legal reviewers ask about contract terms, SLA guarantees, and liability provisions — information that is typically locked behind a sales conversation. Technical evaluators ask about API rate limits, webhook support, and infrastructure architecture — details that live in developer documentation that is often blocked from AI crawlers by robots.txt configurations designed for a pre-AI era.

The enterprise AEO content audit is therefore a two-part exercise: first, identify the questions each buying committee role will ask; second, ensure that the answers to those questions exist as structured, publicly accessible, AI-extractable content on the brand's domain.


THE ENTITY AND AUTHORITY SIGNALS THAT MATTER FOR ENTERPRISE


Enterprise buyers apply a higher credibility threshold to AI-cited information than consumer buyers do. A consumer might act on an AI recommendation without verifying the source. An enterprise buyer — or more precisely, the AI tools they use — will weight citations from sources that carry institutional authority.

This means that the off-site authority signals that matter most in enterprise AEO are different from those that matter in consumer AEO. Consumer AEO benefits from high-volume third-party mentions across a wide range of domains. Enterprise AEO benefits from fewer, higher-authority mentions in the specific sources that enterprise buyers and their AI tools treat as credible: Gartner, Forrester, IDC, Harvard Business Review, industry-specific trade publications, and the websites of major enterprise technology partners.

A single citation in a Gartner Peer Insights report or a Forrester Wave evaluation is worth more for enterprise AI visibility than fifty citations in mid-tier technology blogs. This is not because AI engines explicitly weight Gartner over other sources — though some evidence suggests they do — but because enterprise buyers save and share Gartner content at a far higher rate than they save blog posts. The content that enterprise buyers collect becomes the training data for their internal AI tools. Brands that appear in that content gain a compounding presence inside the enterprise AI ecosystem.


THE MEASUREMENT DIFFERENCE


Consumer AEO measurement is relatively straightforward: track citation rate across public AI engines, monitor AI-referred traffic in analytics, and measure conversion from AI-sourced sessions. These metrics are increasingly available through platforms like Profound, Searchable, and AthenaHQ, and through native tools like Microsoft's Citation Share in Bing Webmaster Tools.

Enterprise AEO measurement is harder, because the most important citation surfaces — internal enterprise AI tools — are not publicly auditable. A brand cannot directly measure how often it is cited by ChatGPT Business Search inside a prospect's Microsoft 365 environment. What it can measure are the proxies: analyst report mentions, G2 review volume and recency, third-party citation rate in the sources that enterprise buyers are known to collect, and pipeline attribution from AI-referred sessions that show the behavioural signatures of enterprise evaluation (multiple page visits, documentation access, security page views, case study downloads).

The most sophisticated enterprise AEO teams are building custom measurement frameworks that combine public AI citation tracking with CRM data to identify the correlation between AI visibility improvements and pipeline velocity. When a brand increases its citation rate in Perplexity responses to enterprise evaluation queries, does the average time-to-close for deals in that category decrease? Does the win rate against specific competitors improve? These are the questions that connect enterprise AEO investment to revenue outcomes in a way that CFOs can evaluate.


THE PRACTICAL STARTING POINT


For most B2B brands selling into enterprise accounts, the enterprise AEO opportunity is largely untapped. The public-web AEO work — robots.txt review, schema implementation, content restructuring for extractability — is the same as for any other AEO programme and should be done first. But the enterprise-specific layer requires three additional actions.

First, audit the content that enterprise buyers actually save and share. Use sales intelligence tools, win/loss interview data, and CRM notes to identify the specific documents, reports, and pages that appear most frequently in enterprise evaluation processes. Ensure that every piece of content in that set is structured for AI extraction: direct answers at the top of each section, named statistics with sources, clear entity signals, and no AI crawler blocks.

Second, build a systematic analyst relations programme with AEO in mind. Every Gartner, Forrester, or IDC mention is an AEO asset. Every G2 review is a citation that will be retrieved by enterprise AI tools. Treat analyst and review platform presence as infrastructure, not as sales collateral.

Third, audit the enterprise AI tool surfaces directly. Create a test environment using ChatGPT Business Search or Microsoft 365 Copilot, populate it with the documents a typical enterprise buyer would collect during an evaluation of your category, and ask the questions each buying committee role would ask. The answers that come back are a direct measure of your enterprise AEO performance — and the gaps they reveal are the most actionable brief an enterprise content team can receive.

The enterprise AEO opportunity is significant precisely because most brands have not yet addressed it. The brands that build this capability now will have a structural advantage in the AI-mediated buying processes that will define enterprise sales in 2027 and beyond.

ChatGPT's new Business Search mode connects internal knowledge bases and CRM data to its answer engine, expanding the citation battle into the enterprise tools buyers use every day.

ChatGPT's new Business Search mode connects internal knowledge bases and CRM data to its answer engine, expanding the citation battle into the enterprise tools buyers use every day.

OpenAI has extended ChatGPT's enterprise tier with a Business Search mode that allows teams to connect internal knowledge bases, CRM records, and document repositories directly to ChatGPT's answer engine. The result is a single interface where employees can ask a question and receive a synthesized answer drawing on both the open web and the company's own proprietary data — with citations surfaced for both sources.

The feature is available to ChatGPT Enterprise and ChatGPT Team subscribers via connectors to Google Drive, Microsoft SharePoint, Salesforce, Notion, and Confluence, with additional integrations in development. Retrieval is permission-aware, meaning users only receive answers grounded in documents they are authorised to access.


WHY THIS MATTERS BEYOND THE ENTERPRISE FIREWALL


The immediate use case is internal productivity: employees get faster, more accurate answers to operational questions without switching between tools. But the AEO implication is more significant than it first appears. Business Search changes the surface on which brand citations occur.

Until now, AEO has been primarily a public-web discipline. Brands optimise content on their own domains, earn citations in third-party publications, and build structured data so that public AI engines can retrieve and cite them accurately. Business Search introduces a new citation surface: the internal AI answer engine used by the enterprise buyers who are evaluating, purchasing, and recommending products. When a procurement team asks ChatGPT Business Search to summarise the leading vendors in a category, the answer it generates draws on whatever content about those vendors exists in the company's connected documents — analyst reports, saved web pages, internal memos, and any public web content the retrieval layer surfaces alongside them.


THE IMPLICATION FOR B2B BRAND VISIBILITY


For B2B brands, this creates a new visibility challenge that sits upstream of the traditional sales cycle. A brand that is well-represented in public AI answer engines but poorly structured in the documents that enterprise buyers save and share — analyst reports, comparison pages, case studies, RFP templates — will underperform in Business Search relative to its public-web citation rate.

The practical response is an extension of existing AEO practice: ensure that the content most likely to be saved and shared by enterprise buyers — detailed product pages, comparison guides, third-party reviews, and case studies — is structured for AI extraction, with clear entity signals, direct answers to common evaluation questions, and authoritative sourcing. Content that is easy for a human to save is also easy for Business Search to retrieve.

OpenAI has not disclosed the number of Enterprise and Team subscribers using Business Search, but ChatGPT Enterprise is reported to have over 600,000 active enterprise users across more than 100,000 organisations. For B2B brands selling into those organisations, Business Search is not a future consideration. It is a current citation surface.

Claude.ai now cites sources inline on factual claims by default, expanding the pool of AI surfaces where brand citations are trackable and attributable.

Claude.ai now cites sources inline on factual claims by default, expanding the pool of AI surfaces where brand citations are trackable and attributable.

Anthropic has updated Claude's default behavior on Claude.ai to include inline source citations on factual claims, bringing it in line with Perplexity and ChatGPT Search. The change, rolled out in a quiet product update, means that Claude now surfaces the web sources it draws on when answering factual queries — displaying them as numbered references within the response body rather than as a separate source list.

The update applies to Claude's web search mode, which uses real-time retrieval to ground responses in current web content. It does not affect Claude's base conversational mode, which draws on training data without live retrieval. For AEO practitioners, the distinction matters: the citation behavior that affects brand visibility is the retrieval-augmented mode, and that is the mode that has now adopted default inline citations.


WHY THIS EXPANDS THE AEO OPPORTUNITY


Until this update, Claude was the major AI answer engine with the least transparent citation behavior. ChatGPT Search and Perplexity both display sources prominently; Google AI Overviews and Bing Copilot surface citations in sidebars. Claude's previous behavior — providing answers without consistently surfacing sources — made it difficult for brands to track whether they were being cited, and difficult for users to verify the provenance of the information they received.

Default inline citations change both dynamics. For brands, Claude's citation behavior is now auditable: AEO platforms can query Claude in retrieval mode and track which sources are cited, in the same way they already track Perplexity and ChatGPT Search citations. AEO platforms are already reporting a spike in Claude citation data requests from enterprise clients following the announcement.


THE PRACTICAL IMPLICATION


Claude has an estimated 50–80 million monthly active users as of mid-2026, with a user base that skews toward professional and enterprise contexts — writers, analysts, developers, and knowledge workers who use Claude for research and synthesis tasks. These are high-value users for B2B brands. A brand that is cited in Claude's response to a research query is reaching a user who is actively building knowledge about a category, often at an early stage of a purchase or vendor evaluation process.

The practical implication for AEO strategy is that Claude should now be treated as a first-tier citation target alongside Perplexity and ChatGPT Search. Content that is structured for extraction, with direct answers, clear entity signals, and authoritative sourcing, is more likely to be retrieved and cited. Brands that have not yet audited their crawlability for ClaudeBot — Anthropic's web crawler — should do so as a priority.

A new Gartner forecast projects that Ask Engine Optimization will graduate from experimental to essential within 18 months, driven by the measurable conversion premium of AI-referred traffic.

A new Gartner forecast projects that Ask Engine Optimization will graduate from experimental to essential within 18 months, driven by the measurable conversion premium of AI-referred traffic.

A new Gartner report on AI-driven marketing investment projects that Ask Engine Optimization will graduate from an experimental line item to a standard budget category within 18 months, appearing alongside SEO and paid search in the majority of enterprise marketing plans by the end of 2027.

The report, part of Gartner's ongoing Digital Marketing Benchmark series, cites the measurable conversion premium of AI-referred traffic as the primary driver of CFO-level buy-in. According to Gartner's analysis, AI search traffic converts at a 3–5x premium over traditional organic search across B2B categories, a differential that finance teams can model directly into revenue forecasts. That modellability — the ability to attach a dollar figure to AI visibility — is what Gartner identifies as the catalyst for AEO's transition from marketing experiment to budget line.


THE MEASUREMENT THRESHOLD


Gartner's forecast is conditional on the maturation of AI visibility measurement infrastructure. The report notes that the two largest barriers to AEO budget adoption are the absence of standardized metrics and the lack of attribution models that connect AI citations to pipeline and revenue. Both barriers are eroding rapidly. Microsoft's Citation Share in Bing Webmaster Tools, Google's AI performance reports in Search Console, and the growing dataset depth of platforms like Profound and Searchable are providing the measurement layer that CFOs require before committing budget.

The report identifies a tipping point: once a brand can demonstrate that a specific piece of AEO-optimized content generated a measurable increase in AI-referred revenue, the internal case for AEO investment becomes self-sustaining. Gartner estimates that approximately 22% of enterprise marketing teams have reached this tipping point as of mid-2026, up from less than 5% in 2025.


IMPLICATIONS FOR THE AEO VENDOR LANDSCAPE


The forecast has direct implications for the AEO platform and agency market. If 60% of enterprise marketing plans include an AEO budget line by end of 2027, the addressable market for AEO platforms and services will expand by an order of magnitude from its current size. Gartner's analysis suggests that the consolidation phase of the AEO vendor market — currently in early proliferation, with over 150 products in the G2 AEO category — will accelerate as enterprise procurement teams seek to standardize on a small number of certified, enterprise-grade platforms.

For brands, the practical implication is straightforward: the window for establishing AEO as an internal competency before it becomes a standard expectation is closing. The brands that build measurement practices, content workflows, and vendor relationships now will enter the consolidation phase with a structural advantage over those that wait for the category to mature.

Perplexity's new Brand Pages let companies verify and manage how they appear inside AI-generated responses, giving brands a direct lever on their AI visibility for the first time.

Perplexity's new Brand Pages let companies verify and manage how they appear inside AI-generated responses, giving brands a direct lever on their AI visibility for the first time.

Perplexity has begun rolling out Brand Pages, a structured profile feature that allows companies to claim and verify their presence inside the platform's AI-generated answers. The feature is currently in limited access for select enterprise partners, with broader availability expected in Q3 2026.

Brand Pages function similarly to Google's Knowledge Panels in intent but are native to the answer engine layer. When a user asks Perplexity about a brand — its products, pricing, use cases, or competitive positioning — the platform can now surface a verified, brand-controlled profile alongside its synthesized response. The profile includes company description, key product information, official website, and a verification badge that signals to users the information has been confirmed by the brand itself.


WHY THIS MATTERS FOR AEO


The launch represents a structural shift in how brands can interact with AI answer engines. Until now, AEO has been an indirect discipline: brands optimize their content, structured data, and off-site authority in the hope that AI systems will cite them accurately. Brand Pages introduce a direct mechanism — a verified source that Perplexity can draw on when generating answers about a company.

The practical implication is significant. Perplexity's citation model is already more transparent than most AI engines, displaying the sources it used to generate each answer. Brand Pages add a new source type to that model: first-party, verified brand data. For brands that have struggled with hallucinations — AI-generated inaccuracies about their products, pricing, or positioning — Brand Pages offer a correction mechanism that does not require waiting for AI training data to update.


THE KNOWLEDGE PANEL PARALLEL


The closest analogy in traditional search is Google's Knowledge Panel, which allows brands to claim their entity in Google's Knowledge Graph and correct inaccurate information. Knowledge Panels became a standard part of enterprise SEO practice within two years of their introduction. AEO practitioners expect Brand Pages to follow a similar adoption curve: initially a differentiator for early movers, then a baseline expectation for any brand with a serious AI visibility strategy.

Perplexity crossed 100 million monthly active users in early 2026. Its user base skews toward high-intent, research-oriented queries — exactly the population that matters most for B2B and considered-purchase brands. A verified Brand Page on Perplexity is not a vanity feature. It is a direct line into the answers that high-value prospects receive when they research a category.

Early access is available to enterprise partners through Perplexity's business development team. AEO Updates will publish a full analysis of the feature's citation mechanics once broader access is available.

Every story we have published in the past six months points in the same direction. Here is the hypothesis, grounded in data, about what AEO looks like in 18 months.

Every story we have published in the past six months points in the same direction. Here is the hypothesis, grounded in data, about what AEO looks like in 18 months.

Journalism is supposed to report what happened. But the stories we have covered over the past six months, taken together, describe something more than a series of product launches and funding rounds. They describe a structural shift in how brand discovery works, and that shift has a direction. This article is an attempt to name it.

The hypothesis is this: AEO is not a new discipline sitting alongside SEO. It is the replacement layer for the entire search-based marketing stack, and the replacement is happening faster than most marketing organisations are built to respond to. The evidence for that hypothesis is sitting inside the news cycle. Here is what it shows.


PATTERN ONE: EVERY MAJOR PLATFORM IS REMOVING FRICTION FROM AI SEARCH


In the past twelve months, every major AI search platform has made a significant move to expand its accessible user base. OpenAI dropped the login requirement for ChatGPT Search in February 2025, making it accessible to anyone with a browser. Google launched AI Mode at Google I/O in May 2026 and crossed one billion monthly active users within weeks. Google AI Overviews now reach two billion users across 200 countries. Microsoft added Citation Share to Bing Webmaster Tools in June 2026, giving publishers a native measurement tool for the first time. Perplexity crossed 100 million monthly users.

These are not incremental feature updates. Each one represents a deliberate expansion of the population of people who receive AI-generated answers instead of a list of links. The login wall coming down at ChatGPT was the most consequential of these moves: it extended AI search to the same casual, intent-driven user who previously typed queries into Google. The implication is that the addressable audience for AI search is now essentially the same as the addressable audience for traditional search. The transition is not coming. It is here.


PATTERN TWO: THE PLATFORMS ARE BUILDING MEASUREMENT INFRASTRUCTURE BEFORE BRANDS ARE READY


A striking feature of the 2026 news cycle is that the platforms are ahead of the brands on measurement. Microsoft launched its AI Performance report in February 2026, expanded it in March, and added Citation Share, Intents, Topics, and Compare in June. Google launched dedicated AI performance reports in Search Console in June 2026. Profound built a database of 1.5 billion real user prompts and now runs over 6 million prompts daily. Searchable tracks nine AI engines and shipped an agentic optimisation layer. AthenaHQ built competitive benchmarking at the query level.

The problem is that only 14% of marketers currently track AI citations, according to research published in 2026. The platforms are building the instruments. The brands have not yet learned to read them. This gap is temporary, but the brands that close it first will have a structural advantage: they will be optimising based on data while their competitors are still debating whether AI search is real.

The measurement infrastructure being built right now will look, in three years, like the early days of Google Analytics. The brands that adopted analytics early did not just measure better. They made different decisions, faster, and compounded those decisions into durable competitive advantages. The same dynamic is beginning in AEO.


PATTERN THREE: CITATION QUALITY IS SEPARATING FROM CITATION QUANTITY


The early AEO conversation was dominated by a simple question: is my brand being mentioned in AI answers? That question is being replaced by a harder one: when my brand is mentioned, does it convert? The data on this is striking. According to Conductor's 2026 benchmarks, AI search traffic converts at 14.2%, compared to 2.8% for Google organic, a five-times premium. A Seer Interactive case study found ChatGPT referral traffic converting at 16%, compared to 1.8% for Google organic, a nine-times premium. The top 2% of brands by AI citation rate are projected to capture $4.2 billion in e-commerce revenue in 2026.

These numbers suggest that AI citations are not just a visibility metric. They are a revenue mechanism. A brand that is cited in response to a high-intent query, in a context that positions it as the authoritative answer, is capturing a user at the moment of maximum purchase readiness. That is a fundamentally different kind of traffic than a user who clicks a blue link from a list of ten results. The implication is that the brands investing in AEO now are not just buying insurance against a future where AI search dominates. They are accessing a conversion channel that already outperforms everything else in the marketing stack.


PATTERN FOUR: THE AGENTIC LAYER IS COMING, AND IT CHANGES EVERYTHING AGAIN


The four patterns above describe the current state of AI search: a human asks a question, an AI answers it, and a brand either appears in that answer or does not. That model is already being disrupted by the next one. Gartner predicts that 40% of enterprise applications will embed AI agents by the end of 2026, up from less than 5% in 2025. Google has integrated agentic capabilities into AI Mode. OpenAI's Operator agent can complete tasks, including purchases, without a human in the loop. 73% of consumers already use AI in their shopping journey.

When the searcher is an AI agent rather than a human, the rules of brand visibility change in a specific way. An agent does not browse. It does not click links and compare options. It queries its training data and its retrieval layer, identifies the most authoritative answer, and acts on it. A brand that is not in the agent's knowledge base, or that is not structured in a way the agent can parse and trust, is not considered. The agent does not know it exists.

This is the end state that the current AEO news cycle is pointing toward: a world where a significant share of purchase decisions are made by AI agents on behalf of human users, and where brand visibility in that world depends entirely on whether the brand has built the kind of authoritative, structured, citation-worthy presence that AI systems recognise as trustworthy. The brands building that presence now are not just preparing for a future state. They are building the infrastructure that will determine whether they exist in the agentic economy at all.


THE HYPOTHESIS, STATED PLAINLY


Here is the prognostication, grounded in the patterns above. In 18 months, the AEO market will look like this: the measurement gap will have closed, and brands that are not tracking Citation Share, AI mention rate, and share of voice across AI engines will be visibly behind their competitors. The platform consolidation that is beginning now will have produced two or three dominant AEO platforms, with the holding groups having acquired or built their own versions. The conversion premium of AI-sourced traffic will have become widely understood, and AEO budget lines will appear in standard marketing plans alongside SEO and paid search. And the first wave of agentic AI shopping experiences will have gone mainstream, making the brands that did not build structured, citation-ready content in 2025 and 2026 invisible to a growing share of purchase decisions.

The window for early-mover advantage in AEO is not closed. But it is closing. The patterns in the news cycle are not ambiguous about the direction. The only question is the pace.

On February 5, 2025, OpenAI made ChatGPT Search available without an account. Eighteen months later, it is the most consequential distribution decision in AI search history.

On February 5, 2025, OpenAI made ChatGPT Search available without an account. Eighteen months later, it is the most consequential distribution decision in AI search history.

On February 5, 2025, OpenAI announced on X that ChatGPT Search was available to anyone, anywhere, without an account. The feature had launched in October 2024 for paying subscribers, expanded to all logged-in users in December 2024, and was now fully open. No signup. No subscription. No login. Just a search bar at chatgpt.com that answered questions with cited sources from the web.

The decision was quiet. There was no press conference, no keynote, no product launch event. But its implications for brand visibility were larger than any of the platform announcements that surrounded it. For the first time, an AI answer engine was accessible to the same population of users as Google: anyone with a browser.


THE SCALE EIGHTEEN MONTHS LATER


By February 2026, ChatGPT had crossed 900 million weekly active users, up from 500 million at the start of 2025. OpenAI reported the figure alongside a new $110 billion funding round. First Page Sage's Q2 2026 estimate placed Google at approximately 80% of all digital queries and ChatGPT at approximately 17%, with the remainder split across Perplexity, Bing Copilot, and other AI platforms. Digital Applied's analysis estimated that ChatGPT Search processes between 250 and 500 million weekly queries, a figure that has grown consistently since the login wall came down.

Outbound referral traffic from ChatGPT to the rest of the web grew 206% in 2025, according to Semrush's 17-month analysis published in April 2026. The volume remains small as a share of total web traffic, with AI referrals accounting for approximately 1.08% of all website visits. But the quality of that traffic is exceptional. According to a Seer Interactive case study, traffic referred from ChatGPT converts at 16%, compared to 1.8% for Google organic, a nine-times premium. Conductor's 2026 benchmarks put the figure at a five-times premium, still the highest conversion rate of any major traffic source.


WHY NO LOGIN CHANGES EVERYTHING


The removal of the login requirement matters for brand visibility in a specific way. When ChatGPT Search required an account, the population of users was self-selected: early adopters, technology professionals, and AI enthusiasts. That population was valuable but narrow. Without a login requirement, ChatGPT Search is now accessible to the same casual, intent-driven searchers who use Google to research purchases, compare providers, and find answers to questions they would previously have typed into a search box.

Those users are the ones whose queries matter most for brand visibility. A user researching 'best CRM for small business' or 'what is AEO' or 'how to improve AI search ranking' is a high-intent user at an early stage of a purchase or learning journey. If ChatGPT Search answers that query without citing a brand, that brand does not exist for that user in that moment. The login wall, while it existed, limited the scale of that risk. Its removal made it universal.


WHAT BRANDS SHOULD DO


The practical implication is straightforward. ChatGPT Search uses a fine-tuned version of GPT-4o, post-trained on web content, and cites sources in a sidebar that mirrors the citation model used by Perplexity and Google AI Overviews. Content that is structured for extraction, with direct answers front-loaded, self-contained answer blocks, and clear entity signals, is more likely to be cited. Any website can appear in ChatGPT Search results by ensuring its content is crawlable by OAI-SearchBot, OpenAI's web crawler. Brands that have not yet audited their crawlability for OAI-SearchBot are invisible to ChatGPT Search by default, regardless of how well they rank on Google.

On June 16, 2026, Microsoft introduced four new AI visibility capabilities in Bing Webmaster Tools. Citation Share measures the percentage of AI responses that reference a brand's content for any given query.

On June 16, 2026, Microsoft introduced four new AI visibility capabilities in Bing Webmaster Tools. Citation Share measures the percentage of AI responses that reference a brand's content for any given query.

On June 16, 2026, Microsoft introduced four new AI visibility capabilities in Bing Webmaster Tools: Intents, Topics, Citation Share, and Compare. The update, available for free in global preview, represents the most significant expansion of Bing's publisher tools since the launch of the original AI Performance report in February 2026 and gives brands a direct, native window into how their content performs inside Microsoft Copilot, Bing, and partner AI experiences.


WHAT CITATION SHARE MEASURES


Citation Share is the most consequential of the four new capabilities. The metric is calculated as a brand's citations divided by total citations for a specific grounding query, expressed as a percentage. If a brand's content is cited in 30 out of 100 AI-generated responses for the query 'best project management software,' its Citation Share for that query is 30%. The metric is the first native measurement from any major platform that directly quantifies a brand's proportional presence in AI-generated answers, rather than simply counting total citations.

The Intents feature surfaces the underlying user intent behind queries that trigger AI responses, allowing brands to understand not just which queries they are being cited for, but what users were trying to accomplish when they asked. Topics groups related queries into thematic clusters, making it easier to identify content gaps at the category level rather than the individual keyword level. Compare allows brands to benchmark their Citation Share against competitors for the same queries.


THE FEBRUARY FOUNDATION


The June update builds on the AI Performance report Microsoft launched in public preview on February 10, 2026. That report provided the first native data on content appearance in AI-generated answers across Microsoft Copilot, Bing, and partner AI experiences, including total citations, average cited pages, grounding queries, and page-level citation activity. The March 23, 2026 expansion of the AI Performance dashboard added advertiser-specific insights into citation frequency in generative answers.


THE MARKET CONTEXT


Microsoft's investment in publisher-facing AI visibility tools reflects the broader commercial stakes of AI search. According to McKinsey, half of consumers are already using AI-powered search, and AI search is projected to influence $750 billion in revenue by 2028. Bing Copilot, while a distant second to Google in overall search volume, is the dominant AI search surface in enterprise and B2B contexts, where Microsoft's deep integration with Office 365, Teams, and Azure gives Copilot a structural advantage over Google's consumer-first AI Mode. For B2B brands in particular, Citation Share in Bing Webmaster Tools may be the single most actionable new metric available in 2026.

Google's AI Overviews now appear on nearly half of all search queries globally. Two new capabilities, image generation and Preferred Sources, have raised the stakes for brands significantly.

Google's AI Overviews now appear on nearly half of all search queries globally. Two new capabilities, image generation and Preferred Sources, have raised the stakes for brands significantly.

Google AI Overviews, the AI-generated summary that appears above traditional search results, has reached 2 billion monthly users across 200+ countries. The feature, which launched in limited form in 2024, now appears on roughly 47 to 64% of all search queries globally, up from 25 to 30% at launch. For brands, the expansion of AI Overviews is not a future consideration. It is the current state of search.


IMAGE GENERATION ARRIVES


On July 14, 2026, Google began rolling out image generation inside AI Overviews. The feature, initially available in English-speaking markets, allows AI Overviews to generate custom visuals in response to user queries, not just synthesise text from existing sources. The rollout has significant implications for brands with strong visual identities: AI-generated images that appear in response to brand-adjacent queries may not reflect brand guidelines, and there is currently no mechanism for brands to influence or correct AI-generated visual representations of their products or services.


PREFERRED SOURCES: AUDIENCE LOYALTY AS A VISIBILITY SIGNAL


In June 2026, Google expanded its Preferred Sources feature from Top Stories into AI Overviews and AI Mode responses. Users can now designate specific publications as preferred, and those preferences follow them into AI-generated answers. More than 345,000 sources have been selected so far, up from approximately 90,000 at the December 2025 global rollout. According to Google, users are twice as likely to click through to a Preferred Source when it appears in an AI response.

Geertrui Laleman, Senior AI Search Optimisation Specialist at Semrush, described the implication directly: 'AI visibility is no longer only about being cited. It is about becoming a source people recognise, trust, and actively want to see.' For brands with existing audiences, the Preferred Sources mechanism creates a direct conversion path from audience loyalty to AI visibility. Newsletter subscribers, social followers, and returning readers are the most likely to designate a brand as a Preferred Source, and each selection increases the probability of that brand appearing in AI responses for that user.


THE NEW SEARCH CONSOLE REPORTS


On June 3, 2026, Google launched dedicated AI performance reports in Search Console, giving site owners for the first time a direct view into how their pages perform inside AI features. The reports show impressions inside AI Overviews, AI Mode, and generative Discover features, broken down by page, country, device, and date. For brands that have been flying blind on AI visibility, the Search Console reports provide the baseline data needed to begin systematic optimisation.

The three leading AI visibility platforms have each shipped significant capability updates in 2026, moving from dashboards to active optimization agents.

The three leading AI visibility platforms have each shipped significant capability updates in 2026, moving from dashboards to active optimization agents.

The AEO platform market has entered a second phase. The first phase, which ran roughly from 2023 to early 2025, was defined by monitoring: platforms competed on how many AI engines they tracked, how many prompts they ran, and how quickly they could surface citation data. The second phase, now underway, is defined by action. The leading platforms are shipping features that do not just measure AI visibility, but improve it.


PROFOUND: FROM ANALYTICS TO AGENTIC WORKFLOWS


Profound, which raised a $35M Series B led by Sequoia in August 2025, has expanded its platform significantly in 2026. The company now tracks 10+ AI engines using a database of 1.5 billion real user prompts, running over 6 million prompts daily. Its most significant 2026 additions are agentic workflows that move beyond reporting: pre-publication content optimisation tools that score content for citation eligibility before it goes live, and a Prompt Volumes feature that provides weekly data on what users are actually asking AI engines. Profound's tracking data on Google AI Mode, which showed google.com becoming the second most-cited domain inside AI Mode, has been widely cited across the industry as a benchmark data source.


SEARCHABLE: THE AI AGENT LAYER


Searchable, the London-based platform that raised $14M at an $85M valuation in May 2026, has shipped what it describes as an AI agent layer on top of its monitoring infrastructure. The agent goes beyond dashboards to generate actionable recommendations, content briefs, and technical fixes based on citation gap analysis. Searchable now tracks 9 AI engines, including ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, Copilot, Grok, Meta AI, and DeepSeek. Its AI Search Accelerator training programme has enrolled more than 1,000 brands. The platform has achieved G2 High Performer status for Summer 2026.


ATHENAHQ: THE CHALLENGER


AthenaHQ has emerged as the most aggressive challenger in the platform market in 2026. The platform's positioning centres on brand share of voice across AI engines, with a particular focus on competitive benchmarking at the query level. AthenaHQ tracks which competitors are being cited for the same queries a brand is targeting, and surfaces the content and structural differences that explain the citation gap. The platform has attracted significant attention from enterprise marketing teams that want competitive intelligence alongside their own visibility data.


THE MARKET SIGNAL


The broader market context makes the platform race urgent. According to Conductor's 2026 benchmarks, AI search traffic converts at 14.2%, compared to 2.8% for Google organic, a five-times premium. Google AI Overviews now appear on 48% of all search queries as of March 2026. Organic click-through rates at position one drop by approximately 34.5% when an AI Overview is present. The platforms that can move brands from monitoring to active citation improvement are now the most valuable tools in the marketing stack, and the 2026 feature drops from Profound, Searchable, and AthenaHQ are the clearest evidence yet that the category has reached that capability threshold.

Launched at Google I/O on May 19, 2026, AI Mode has become the fastest-growing search surface in Google's history. For brands, the implications are immediate.

Launched at Google I/O on May 19, 2026, AI Mode has become the fastest-growing search surface in Google's history. For brands, the implications are immediate.

On May 19, 2026, at Google I/O, Liz Reid, Google's VP of Search, confirmed that AI Mode had crossed one billion monthly active users. Queries inside the interface were doubling every quarter. The announcement marked a structural inflection point: for the first time in the history of commercial search, the dominant search experience returns zero blue links.

AI Mode is not a feature layered on top of Google Search. It is a replacement. Powered by Gemini 3.5 Flash, the interface accepts conversational queries, processes follow-up questions, and synthesises multi-source answers without ever presenting a traditional results page. Users who ask a question inside AI Mode receive a synthesised response. Citation, not ranking, is the only visibility mechanism available to brands.


THE NUMBERS THAT MATTER


The scale of the rollout is difficult to overstate. AI Mode is now available in 180+ countries. The March 2026 core update, which ran for 12 days and caused significant ranking volatility, was partly a recalibration of Google's quality signals to support AI Mode's citation model. By June 2026, Profound's tracking data showed that google.com had become the second most-cited domain inside AI Mode, driven almost entirely by Google Business Profiles and Product Knowledge Panels appearing as inline panels within AI responses.

That last point deserves attention. Profound tracked AI Mode citation share from April 15 through June 30, analysing more than 32 million instances. The increase in google.com citations came from Google-hosted cards, not from google.com editorial content. For local and product searches, the Google Business Profile is now the first page many users see, appearing before any brand website is ever reached.


WHAT CHANGED ON JULY 16


On July 16, 2026, Google expanded AI Mode further by adding Connected Apps: integrations with Instacart, Canva, and YouTube Music that allow users to complete tasks directly inside AI Mode without leaving the search interface. A user can now ask AI Mode to build a shopping list, generate a design brief, or queue a playlist without a single click to an external site. The connected apps expansion is the clearest signal yet that Google intends AI Mode to become a task-completion layer, not just an information layer.


THE BRAND IMPERATIVE


For brands, the shift demands a reorientation of search strategy. Traditional organic rankings remain relevant, as 92% of AI Overview citations still come from top-10 ranking pages. But citation eligibility requires more than ranking: content must be structured for extraction, with direct answers front-loaded in each section, self-contained answer blocks of 134 to 167 words, and Article and FAQPage schema markup applied. Google Business Profiles must be treated as primary brand assets, with hours, photos, and reviews maintained to the same standard as a brand's own website. And share of voice inside AI Mode, not click-through rate, is the metric that will define brand visibility for the next decade.

The world's largest agency groups are quietly building AI search visibility into their proprietary platforms. None of them have launched a named AEO practice. That gap is the story.

The world's largest agency groups are quietly building AI search visibility into their proprietary platforms. None of them have launched a named AEO practice. That gap is the story.

Ask any of the six major agency holding groups whether they offer AEO services and you will get a carefully worded non-answer. Ask them whether their proprietary AI platforms include tools to monitor and improve how client brands appear in large language models, and every single one will say yes. That gap between the capability and the category name is the most revealing thing happening in the AEO industry right now.


THE PLATFORM LAYER


In April 2026, The Media Leader published a comprehensive breakdown of every holding group's AI platform. The findings were striking not for what the platforms do, but for how they are positioned. WPP's Open platform, launched in 2024 as a 'large marketing model,' includes an Agent Hub with tools for AI search monitoring. Publicis's Marcel — the oldest of the group platforms, launched in 2018 — has been integrated with Microsoft Azure and now includes Epsilon identity data for AI-driven audience targeting. Havas's Converged.AI explicitly includes an LLM brand monitoring module. Omnicom's Omni, relaunched after the IPG acquisition, includes 2.6 billion verified IDs and real-time commerce signals that feed into AI-driven campaign optimization. Dentsu's Connect platform, revamped in 2026, incorporates Google Gemini and Meta Llama for multi-agent workflows.

What none of these platforms do is offer a standalone, named AEO service. There is no 'WPP AEO.' There is no 'Publicis Answer Engine Practice.' The capability exists, embedded inside broader AI marketing platforms, but the category framing — the explicit acknowledgment that AI search visibility is a distinct discipline requiring distinct methodology — is absent.


THE ONE EXCEPTION: RAZORFISH


The exception is Razorfish, the Publicis-owned digital agency. In September 2025, Razorfish hosted a client-facing webinar titled 'The AI Search Playbook: Breakthrough Brand Visibility,' led by Amos Ductan, SVP of Search. The event explicitly addressed how AI platforms rank and surface brand content, what leading marketers are doing to adapt, and the risks of waiting too long to act. It is the only public-facing, named AI search optimization offering from any holding group entity — and it came from a specialist digital agency within the network, not from the holding group itself.

That distinction matters. Razorfish's AI Search Playbook is a practitioner-level offering built by people who run search programs. The holding group platforms are infrastructure plays built by people who run data and technology businesses. They are solving different problems for different buyers, and the fact that Razorfish had to build its own offering rather than leverage WPP Open or Marcel suggests that the platform layer is not yet close enough to the practitioner layer to serve client needs directly.


WHAT THE HOLDCOS ARE ACTUALLY DOING


The more accurate description of what the holding groups are doing on AEO is monitoring, not optimization. Every major group now has the capability to track how client brands appear across ChatGPT, Gemini, Perplexity, and Copilot. Havas's Brand Insights tool, Omni's commerce signals layer, and WPP Open's Agent Hub all include some version of LLM brand monitoring. What they do not yet have — at least not in any publicly documented form — is a systematic methodology for improving those appearances: the content restructuring, entity building, third-party citation strategy, and technical crawlability work that constitutes actual AEO practice.

That gap is not a criticism. It reflects the natural lag between a new category emerging and large organizations building the internal capability to serve it. The SEO analogy is instructive: when Google's PageRank algorithm became the dominant discovery mechanism in the early 2000s, the holding groups did not immediately build SEO practices. They acquired specialist agencies, incubated internal teams, and eventually integrated SEO into their broader digital offerings over a period of years. The AEO cycle is likely to be faster — the category is moving faster, the client demand is more urgent, and the holding groups are more structurally prepared to absorb new digital disciplines than they were in 2002.


THE INDEPENDENT ADVANTAGE


The practical implication for brands is that the most sophisticated AEO capability currently sits outside the holding group networks. Platforms like Profound, Searchable, and Goodie AI have built purpose-built AI visibility infrastructure that the holding group platforms have not yet matched. Full-service AEO agencies have developed optimization methodologies that the holding group monitoring tools do not yet execute. And independent consultants like Kevin Indig and Aleyda Solis are publishing the research and frameworks that the entire industry — including the holding groups — is learning from.

That balance will shift. It always does. But for the next 18 to 24 months, the brands that move fastest on AEO will be the ones that engage directly with the specialist ecosystem rather than waiting for their holding group to build the capability internally. The holdcos are coming. They are just not here yet.

From Sequoia's $35M bet on Profound to Headline's investment in Searchable, institutional capital is moving fast into the AEO space. What the funding patterns reveal about where the market is heading.

From Sequoia's $35M bet on Profound to Headline's investment in Searchable, institutional capital is moving fast into the AEO space. What the funding patterns reveal about where the market is heading.

In the summer of 2024, the AEO market barely existed as a recognized investment category. Twelve months later, Sequoia Capital had led a $35 million Series B into Profound, Kleiner Perkins had anchored a $20 million Series A into the same company, Headline VC had backed Searchable at an $85 million valuation, and a dozen other platforms had raised pre-seed and seed rounds from investors who had spent the previous decade backing SEO and content marketing technology. The total capital deployed into AEO-specific platforms and agencies in an 18-month window now exceeds $150 million. For a category that did not have a name three years ago, that is a remarkable concentration of institutional conviction.


WHY NOW


The investment thesis is not complicated, but it is consequential. AI-powered chatbots now account for more than five percent of U.S. desktop search traffic, up from 1.3 percent in early 2024. ChatGPT crossed one billion weekly active users in 2025. More than 60 percent of consumers now start product research with an AI assistant rather than a traditional search engine. And 65 percent of all searches — across both Google and AI platforms — now end without a click. The implication for brands is stark: the primary discovery channel for a generation of consumers is one over which they currently have almost no visibility, no measurement, and no optimization capability.

That gap between the scale of the problem and the maturity of the tooling is exactly the kind of market condition that attracts institutional capital. Sequoia partner Anas Biad described the firm's thesis as a 'once-in-a-generation platform shift for marketers.' Kleiner Perkins partner Ilya Fushman called Profound 'the clear leader in a category that will define modern marketing.' Headline's Dominic Wilhelm framed it as 'one of the most important customer acquisition channels of the next decade.' These are not hedged, exploratory bets. They are categorical statements of conviction from firms that have collectively backed Google, Airbnb, Stripe, Bumble, and Semrush.


THE FUNDING MAP


The capital has concentrated in a predictable pattern. Pure-play AI visibility platforms have attracted the largest rounds, because they are the most legible investment thesis: a SaaS business with recurring revenue, measurable ROI, and a clear analogy to the SEO software market that Semrush and Ahrefs built into billion-dollar businesses. Profound's $58.5 million total raise is the largest in the category. Searchable's $18 million across two rounds at an $85 million valuation is the fastest trajectory. Goodie AI, Otterly, Peec AI, and AthenaHQ have all raised pre-seed and seed rounds in the $1 to $5 million range from angel investors and early-stage funds.

Full-service AEO agencies have attracted less formal venture capital, for the structural reason that services businesses carry lower multiples than SaaS. But several have raised growth equity or taken on strategic investment. The more interesting dynamic in the agency segment is consolidation: traditional SEO agencies acquiring AEO specialists, and AEO agencies building proprietary platforms to differentiate from pure-play SaaS competitors. That convergence is creating a new category of hybrid platform-agency that does not fit neatly into either the SaaS or services investment thesis.


THE SEMRUSH PARALLEL


The most frequently cited historical analogy among AEO investors is Semrush. Founded in 2008, Semrush spent a decade building the dominant SEO analytics platform before going public in 2021 at a $1.9 billion valuation and subsequently being acquired by Adobe. The parallel is instructive but imperfect. Semrush built its business in a relatively stable search environment where Google's algorithm changed incrementally and the fundamental mechanics of SEO were well understood. AEO platforms are building in an environment where the underlying AI models change weekly, the citation logic is opaque and probabilistic, and the competitive landscape includes not just other AEO platforms but the AI companies themselves — OpenAI, Anthropic, and Google — all of whom have their own incentives around how brands appear in AI-generated answers.

That volatility cuts both ways for investors. It creates execution risk for the platforms: a model update from OpenAI can change citation patterns overnight, and a platform that cannot adapt its tracking and optimization capabilities fast enough will lose relevance quickly. But it also creates a durable moat for the platforms that do adapt: the proprietary prompt databases, citation attribution models, and content optimization frameworks that the leading platforms are building are genuinely difficult to replicate. Profound's 1.5 billion real user prompt database is not something a new entrant can build in six months.


WHAT PRIVATE EQUITY IS WATCHING


Venture capital has moved first, as it always does in emerging categories. Private equity is watching from a closer distance than most observers realize. The conditions that typically trigger PE interest — category definition, revenue predictability, consolidation opportunity, and a clear path to platform-level scale — are beginning to emerge in AEO. The G2 AEO software category grew 2,000 percent in 2025 alone. The 94 percent of CMOs who told Conductor they plan to increase AEO spending in 2026 are not all going to the same type of provider. The brands that navigate this landscape most effectively will be the ones that understand which category of provider solves which category of problem — and resist the temptation to treat AEO as a single, undifferentiated service.

The more likely near-term PE play is not a direct investment in an AEO platform but a roll-up of the agency segment. The AEO agency market is fragmented, founder-led, and operating in a category where brand and methodology differentiation matter enormously. That is a classic PE consolidation setup. A platform that acquires three or four leading AEO agencies, standardizes their methodology, and builds a proprietary technology layer on top of the combined client base would have a credible path to the kind of EBITDA margins and revenue predictability that PE underwriting requires.


THE RISK SCENARIO


No honest assessment of the AEO investment landscape can ignore the risk scenario. The category is being built on the assumption that AI search will continue to grow its share of the discovery journey — an assumption that is well-supported by current data but not guaranteed. Google has demonstrated repeatedly that it can adapt its core product to absorb emerging threats, and its AI Overviews rollout represents a direct move to capture AI search behavior within its own ecosystem. If Google successfully retains its dominant position in the AI search era, the addressable market for third-party AEO platforms narrows considerably.

There is also a commoditization risk. The core functionality of AI visibility monitoring — tracking brand mentions across AI platforms — is not technically complex. Several free or low-cost tools have emerged that offer basic citation tracking. The platforms that will survive and scale are the ones that move up the value stack from monitoring to optimization to execution, building the kind of workflow integration and proprietary data assets that create genuine switching costs. The funding rounds going to Profound and Searchable suggest that investors believe those platforms are on the right trajectory. The rest of the market will need to prove the same.

The capital is in. The category is defined. The question now is which platforms and agencies will build the durable competitive advantages that justify the valuations being assigned today — and which will be acquired, consolidated, or left behind as the market matures.

The New York and London-based platform, backed by Headline VC, is targeting a 40 percent reduction in SEO costs while growing AI-driven traffic for enterprise brands including American Express, KPMG, and Siemens.

The New York and London-based platform, backed by Headline VC, is targeting a 40 percent reduction in SEO costs while growing AI-driven traffic for enterprise brands including American Express, KPMG, and Siemens.

Searchable, the AI performance marketing platform helping businesses compete in AI-driven search, has raised $14 million in funding led by global venture capital firm Headline at an $85 million valuation. The round follows a December 2025 pre-seed of $4 million at a $40 million valuation — meaning the company has nearly doubled its valuation in under six months, a trajectory that reflects the accelerating urgency with which enterprise marketing teams are approaching AI search visibility.

Headline's portfolio includes Bumble, Farfetch, Goop, and Sonos, and notably Semrush — the SEO software group acquired by Adobe in a $1.9 billion deal. The firm's prior investment in Semrush gives it a distinctive vantage point on the transition from traditional SEO to AI-native search optimization, and its conviction that Searchable can occupy a similar category-defining position in the new era.


THE PLATFORM


Founded in 2025 by British serial entrepreneur Chris Donnelly — who previously sold SEO agency Verb for $25 million and scaled Lottie Org to a nine-figure valuation — Searchable tracks visibility across 10 AI engines including ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, Copilot, Grok, Meta AI, and DeepSeek. The platform combines monitoring with an agentic optimization layer that generates content briefs, technical fixes, and actionable recommendations rather than just reporting on current visibility.

The company reached $2 million in revenue in 4.5 months and has onboarded nearly 1,000 customers. Enterprise-scale customers including American Express, KPMG, and Siemens report a 22 percent increase in AI-driven traffic within their first 60 days on the platform. The company holds a G2 High Performer badge for Summer 2026 with a 4.8/5 rating across 93 reviews.


THE INVESTMENT THESIS


Dominic R. Wilhelm, Partner at Headline, framed the investment in terms that go beyond product features. 'AI-driven discovery is rewriting how customers find products. As more searches are answered directly by AI, brands that are invisible in this layer of search will see less demand. The companies that adapt first will grow market share; those that don't will lose it quietly.' Wilhelm added that Headline sees Searchable becoming 'part of the core infrastructure for this shift, not just reporting on what AI engines say about a brand, but directly improving the visibility and revenue outcomes that matter to management teams and boards.'

Donnelly's own framing of the opportunity is grounded in a specific conversion data point his platform has generated: customers arriving from ChatGPT and other LLMs are converting at three times the rate of traditional organic search visitors. If that figure holds at scale, it reframes AI search visibility not as a brand awareness play but as a direct revenue driver — a shift that changes the budget conversation from marketing to growth.


WHAT COMES NEXT


With the new capital, Searchable plans to accelerate product development across its execution engine and expand its presence in both the U.S. and U.K. markets. Over the next 12 to 24 months, Donnelly expects three structural shifts to reshape the market: the automation of repetitive manual SEO labor through agentic systems; the rise of AI commerce as a standalone optimization layer; and the convergence of paid and organic AI visibility into unified attribution models. 'Our goal is to give companies an execution layer for AI search that cuts SEO costs by up to 40 percent while growing high-intent traffic,' he said.

The New York-based AEO platform has now raised $58.5M in total funding, with Sequoia Capital leading its latest round as Fortune 10 brands race to secure visibility in AI-generated answers.

The New York-based AEO platform has now raised $58.5M in total funding, with Sequoia Capital leading its latest round as Fortune 10 brands race to secure visibility in AI-generated answers.

Profound, the AI visibility platform that claims to have been the first company to stake out the post-SEO space, has raised a $35 million Series B funding round led by Sequoia Capital. The round brings Profound's total funding to $58.5 million, following a $3.5 million seed led by Khosla Ventures and a $20 million Series A led by Kleiner Perkins in June 2025. Continued participation from Kleiner Perkins, Khosla Ventures, Saga VC, and South Park Commons signals strong conviction from the company's existing investor base.

The round comes as AI-powered chatbots now account for more than five percent of U.S. desktop search traffic — up from just 1.3 percent in early 2024, according to the Wall Street Journal. Profound says 2,000 marketers from over 500 organizations now use its platform daily, with publicly named clients including Ramp, U.S. Bank, Indeed, MongoDB, DocuSign, and Chime.


WHAT PROFOUND DOES


Profound tracks how major AI models — from ChatGPT, Grok, and Meta's Llama to Google Gemini and Microsoft Copilot — surface brand mentions across more than 1.5 billion real user prompts. The platform processes over 100 million AI search queries each month, supports customers in 18 countries and six languages, and uses advanced reasoning models including OpenAI's o3 to generate content recommendations and gap analyses. Early adopters report a 25 to 40 percent lift in AI answer share-of-voice within 60 days.

The Series A in June 2025 also launched Profound Lite, a self-serve plan at $499 per month aimed at startups and small businesses — a deliberate move to expand the addressable market beyond enterprise. The Series B capital will be used to expand Profound's engineering and data-science teams in New York, deepen coverage across emerging AI platforms, and accelerate product innovation for what co-founder James Cadwallader calls the 'agentic internet.'


SEQUOIA'S THESIS


Sequoia partner Anas Biad told Fortune that the firm views the rise of AI search as a once-in-a-generation platform shift for marketers. 'Their speed of execution was truly remarkable — in both the product they built and the customers they landed,' Biad said. 'We only back founders who want to build generational companies. Their team is very ambitious and very aggressive.'

Cadwallader, who previously founded influencer marketing firm Kyra, has described the shift as a 'Game of Thrones power shift' from decades of SEO tactics to a new world of AI search. His co-founder Dylan Babbs is a former Uber software engineer. The two met at South Park Commons, the San Francisco community and early-stage fund founded by Facebook's first female engineer Ruchi Sanghvi and former Dropbox CTO Aditya Agarwal.


THE BIGGER PICTURE


Cadwallader's ambitions extend well beyond search optimization. He envisions a future where transactions happen directly inside AI assistants — without a single click away — posing competitive threats to giants like Amazon. 'I'm inspired by Salesforce,' he said, pointing to the company's early-2000s cloud software disruption. 'It's an example of just how big you can go.'

The challenge, he acknowledges, has already expanded beyond marketing. 'It's become a PR challenge, a content challenge, even a customer support challenge. The models have opinions, and they reflect the internet's opinions back to users.' For Profound, that breadth of use case is a feature, not a scope problem. The platform is positioning itself not as a search optimization tool but as the foundational infrastructure layer for brand presence in an AI-first internet.

A comprehensive guide to every category of AEO provider — from pure-play visibility platforms to traditional SEO agencies making the pivot.

A comprehensive guide to every category of AEO provider — from pure-play visibility platforms to traditional SEO agencies making the pivot.

Twelve months ago, the AEO vendor market barely existed as a recognized category. Today it is one of the fastest-growing segments in marketing technology, with the G2 AEO software category expanding by 2,000 percent in 2025 alone. That growth has attracted every type of provider imaginable — purpose-built platforms, legacy SEO agencies, boutique consultancies, and individual practitioners — all operating under the same AEO banner but doing fundamentally different things.

Understanding the landscape requires understanding the categories. AEO is not a single service. It is a discipline that spans technical infrastructure, content strategy, entity management, citation monitoring, and off-site authority building. Different providers specialize in different parts of that stack, and the right choice depends entirely on where a brand sits in its AEO maturity and what problem it most urgently needs to solve.

AI Visibility Platforms form the foundation of the market. These are SaaS tools — Profound, Conductor, Searchable, Goodie AI, Otterly AI, and others — that track where a brand appears in AI-generated answers, how often it is cited, and how it compares to competitors across platforms like ChatGPT, Perplexity, Google AI Overviews, and Claude. The best platforms in this category go beyond dashboards to provide actionable optimization guidance and, in some cases, an agentic layer that proactively identifies and fixes visibility gaps. Profound leads the category with an AEO Score of 92/100 and a prompt database of 1.5 billion real user queries. Conductor differentiates by integrating AEO tracking with traditional SEO in a single enterprise platform.

Full-Service AEO Agencies represent the managed services layer of the market. These are organizations — The Prompt Group, Discovered Labs, NoGood, Virayo, and Minuttia among them — that take ownership of the entire AEO operation for a client. They conduct the initial AI visibility audit, map domain territory, deploy structured content engineered for AI citation, implement entity signals, and monitor citation performance on an ongoing basis. The defining characteristic of this category is that they treat AEO as an operational discipline, not a one-time project. Retainer structures are the norm, and the best agencies in this category can demonstrate measurable citation uplift within 30 to 90 days.

Traditional SEO agencies with AEO practices represent the largest and most heterogeneous segment of the market. Firms like Optimist, Omniscient Digital, iPullRank, First Page Sage, and Amsive have built AEO capabilities on top of established SEO foundations. The quality of these practices varies considerably. The best of them — iPullRank's Relevance Engineering framework, First Page Sage's thought leadership model, Optimist's CORE framework — are genuinely differentiated and grounded in how AI systems actually evaluate content. The worst are SEO agencies that have added 'AEO' to their services page without changing their methodology. The critical question to ask any SEO agency claiming AEO expertise is: can you show me a case study with AI-referred revenue or citation uplift metrics, not organic traffic?

Independent consultants occupy a distinct and often underappreciated position in the landscape. Practitioners like Kevin Indig, Aleyda Solís, Lily Ray, Marie Haynes, and Ross Simmonds bring a depth of research and methodological rigor that most agencies cannot match. Indig's correlation analysis of 1.2 million ChatGPT responses, Solís's multilingual GEO framework, and Ray's E-E-A-T diagnostic approach for AI citation gaps are among the most cited bodies of work in the field. For enterprise brands with internal teams that need strategic direction rather than execution, an independent consultant often provides better value than a full-service agency engagement.

Link building and PR specialists form the fifth category, and it is one that is frequently overlooked in AEO discussions. Given that 85 percent of AI citations come from third-party pages, the agencies that specialize in earning editorial placements and brand mentions in authoritative publications are doing some of the most impactful AEO work in the market — even when they do not use the term AEO to describe it. Platforms like Respona and agencies like Dofollow are building the off-site authority signals that AI engines use to evaluate brand credibility.

The market is still young enough that category boundaries are porous. Several platforms are expanding into managed services. Several agencies are building proprietary platforms. The consultants who publish the most cited research are increasingly affiliated with platform companies. What is clear is that the AEO market has moved beyond the early-adopter phase. The 94 percent of CMOs who told Conductor they plan to increase AEO spending in 2026 are not all going to the same type of provider. The brands that navigate this landscape most effectively will be the ones that understand which category of provider solves which category of problem — and resist the temptation to treat AEO as a single, undifferentiated service.

What actually drives citation in AI engines? Synthesizing findings from 50+ published studies to understand the new rules of visibility.

What actually drives citation in AI engines? Synthesizing findings from 50+ published studies to understand the new rules of visibility.

The rules governing which websites and content pieces get cited by AI engines are fundamentally different from the rules that governed traditional search rankings. For two decades, domain authority was the undisputed king of visibility. In the era of AI search, that is no longer the case.

Recent data reveals that domain authority explains only 4 to 7 percent of AI citation outcomes. The remaining 93 to 96 percent is determined by a layered set of technical, structural, content, and off-site factors that most brands have not yet optimized for.

By synthesizing findings from more than 50 published studies — including the Princeton GEO paper, SE Ranking's analysis of 2.3 million pages, Ahrefs' study of 75,000 brands, and Kevin Indig's audit of 1.2 million ChatGPT responses — a clear framework emerges. AI visibility is not a single metric, but a composite of five distinct domains.

Technical Crawlability and Access determines whether AI bots can physically reach and render your content. LLM bots now crawl 3.6 times more frequently than Googlebot, yet 62 percent of news publishers block GPTBot and 69 percent block ClaudeBot, often through outdated robots.txt configurations or CDN defaults. Cloudflare changed its default bot management configuration in 2025 to block AI crawlers automatically — any site that has not reviewed its settings since then may be blocking every AI crawler without knowing it.

Structured Data and Entity Signals give AI systems machine-readable context about what your content represents. Schema-marked pages were cited 2.3 times more often in AI Overviews than comparable unstructured pages. The Knowledge Graph now holds over 1.6 trillion facts about 54 billion entities. Brands without a Wikipedia or Wikidata entry are not viewed as entities — they are treated as strings of text. Strings do not get cited; entities do.

Content Structure and Extractability covers how content is organized on the page. AI systems do not read a page the way a human does. They scan for structure, extract passages from the top of the page first, and prefer content that answers a question directly in the first sentence of each section. 44.2 percent of all LLM citations come from the first 30 percent of a page's text.

Content Quality and Linguistic Signals is where the actual citation decision is made. The Princeton GEO study found that keyword stuffing decreased AI visibility by 10 percent. Adding citations to content boosted AI visibility by 30 to 40 percent, expert quotes lifted it 41 percent, and statistics raised it 30 percent. A single well-sourced sentence with a named study, a specific percentage, and a named population is worth more for AI visibility than three paragraphs of keyword-optimized prose.

Off-Site Authority and Ecosystem Presence accounts for the signals that exist outside the brand's direct control. An astonishing 85 percent of AI citations come from third-party pages, not brand-owned domains. Earned media stories are cited by AI at 239 percent the rate of brand-owned content. PR strategy is now a search strategy.

Why Answer Engine Optimization (AEO) is more important than SEO ever was — and what it means for every CMO making budget decisions today.

Why Answer Engine Optimization (AEO) is more important than SEO ever was — and what it means for every CMO making budget decisions today.

For two decades, Search Engine Optimization was the undisputed king of digital marketing. It was the primary channel for discovery, the first touchpoint for countless customer journeys, and the foundation of online visibility. But the era of SEO as we knew it is over. A new, more powerful force has emerged, and it is not just changing how we find information. It is changing how we form opinions.

That force is Answer Engine Optimization (AEO). It is not just the next evolution of SEO; it is a fundamentally different and more critical marketing discipline. While SEO surfaced brands, AEO shapes perceptions. While SEO provided research, AEO delivers recommendations. And while SEO was a lagging indicator of brand awareness, AEO is a leading indicator of brand perception.

If you are still thinking about online visibility through the lens of SEO, you are not just behind the curve. You are playing a different game entirely. Here is why AEO is the most important marketing investment your company can make today.


THE SHIFT FROM RESEARCH TO RECOMMENDATION


The fundamental difference between SEO and AEO lies in user behavior and intent. When a user performs a traditional Google search, they are presented with a list of options — the infamous ten blue links. This format encourages research and comparison. The user clicks multiple links, reads different sources, and synthesizes information to form their own opinion. In this model, the brand's role is to be present in the consideration set, to be one of the options the user evaluates.

AI-powered answer engines like ChatGPT, Perplexity, and Google's AI Overviews have fundamentally changed this dynamic. When a user asks a question like 'what are the best restaurants in Austin?', they are often presented not with a raw list of links but with a synthesized, pre-evaluated answer. The AI has already processed the options, weighted the sources, and framed the consideration set. The user still decides, but the AI has done the shortlisting. That is a different kind of influence than a list of ten blue links.

This shift from a research-based interaction to a pre-evaluated one is significant. One survey found that 76 percent of Gen Z and younger millennials report trusting AI answers more than traditional Google results. [1] The sample was small and non-representative, so treat the precise figure with appropriate caution. But the directional signal is consistent with what practitioners are observing: a meaningful share of users, particularly younger ones, are treating AI answers as a credible starting point rather than one source among many. That is a change in trust dynamics worth taking seriously.


AEO: THE NEW FRONTIER OF PERCEPTION SHAPING


Because AI-powered answers are treated as recommendations, they shape user perception before a brand ever gets a chance to speak for itself. If an AI recommends your competitor's product, it is not just that your competitor is more visible — it is that they are perceived as superior. The AI has conferred its authority and trust onto that brand, and you are left on the outside looking in.

This is why AEO is so much more important than SEO ever was.

"SEO was about being seen; AEO is about being chosen. SEO was about awareness; AEO is about authority. SEO was about getting on the list; AEO is about being the list."

Alphabet reported $119.8B in revenue and $112.1B in net income for Q2 2026. Beneath the headline is a story about where Google is betting its future: cloud, capex, and the infrastructure layer of AI.

Alphabet reported $119.8B in revenue and $112.1B in net income for Q2 2026. Beneath the headline is a story about where Google is betting its future: cloud, capex, and the infrastructure layer of AI.

On July 22, Alphabet reported the largest quarterly profit in its history. Net income reached $112.1 billion, up 298% year-over-year, on revenue of $119.8 billion. The result included a $99 billion net gain on Alphabet's equity holdings — primarily from SpaceX and a second unnamed private company — which generated $98 billion of net other income and added $6.26 to earnings per share.

Strip out that gain and EPS was roughly $2.85, slightly below the $2.88 to $2.89 analysts expected. The headline number is real money, but it is not operating money. It is a mark-to-market gain on a portfolio bet that happened to mature in a single quarter.

<h2>The Number That Actually Matters: Google Cloud</h2>

The clearest signal of operating strength was Google Cloud. Revenue rose 82% to $24.8 billion. Operating income reached $8.8 billion, producing a 35.6% margin. For context, Google Cloud was losing money as recently as 2022. It is now one of the most profitable cloud businesses in the world, and it is growing faster than AWS or Azure on a percentage basis.

For AEO practitioners, this matters directly. Google Cloud is the infrastructure layer for AI Mode, Gemini, and the answer engine that is reshaping how brands get discovered. A cloud business growing at 82% with 35.6% margins is a business that can fund indefinite investment in AI search infrastructure. The competitive moat is not the model. It is the infrastructure.

<h2>The Capex Signal</h2>

Capital spending roughly doubled to $44.9 billion in the quarter, resulting in free cash flow of negative $5.9 billion — Alphabet's first negative free cash flow quarter. Alphabet also raised its full-year 2026 capex guidance to $195 billion to $205 billion, up from $180 billion to $190 billion.

Negative free cash flow at this scale is a deliberate choice, not a distress signal. Alphabet is spending more than it earns in cash this quarter because it has decided the infrastructure window is now. Every dollar of that capex is going into data centers, TPUs, and the network capacity that AI Mode and Google Cloud require. The companies that win the infrastructure layer of AI will set the terms for everything above it.

<h2>The Anthropic Connection</h2>

Alphabet reportedly owns approximately 14% of Anthropic. During Q2, Anthropic's valuation rose from $380 billion to $965 billion following a $65 billion funding round. That appreciation is almost certainly a significant component of the $99 billion equity gain. The implication is that Alphabet's balance sheet is now partially correlated with the success of one of Google's most direct AI competitors — a structural tension that will become more visible as the AI search market consolidates.

<h2>What This Means for the AEO Landscape</h2>

Three things are now clearer after this quarter. First, Google is not a company in transition — it is a company in acceleration. The infrastructure investment at this scale means AI Mode, AI Overviews, and Gemini integrations will expand faster and more aggressively than most brands are prepared for. Second, Google Cloud's growth rate means enterprise AI adoption is real and accelerating, which expands the surface area where brands need AI visibility beyond consumer search. Third, the negative free cash flow quarter signals that Google has decided the infrastructure race is existential. Brands that treat AEO as a 2027 priority are already behind.

The record quarter is a financial story. The capex guidance and Cloud margin are the AEO story.

<h3>References</h3>

The world's largest marketing platform has built a dedicated AEO product. Here is what it does, what it costs, and what it means for the category.

The world's largest marketing platform has built a dedicated AEO product. Here is what it does, what it costs, and what it means for the category.

HubSpot — the company whose 2025 revenue was $846.7M and whose CRM platform is used by more than 248,000 businesses worldwide — has entered the AEO market. At its Spring 2026 Spotlight event, HubSpot launched two AEO products: a free AEO Grader and a paid HubSpot AEO monitoring tool at $50/month (or $45/month annually), bundled into Marketing Hub Pro and Enterprise.

The free AEO Grader is a one-time brand visibility diagnostic. Enter a URL and a target keyword, and the tool simultaneously queries ChatGPT, Perplexity, and Gemini, returning a score out of 100 across five dimensions: Sentiment (40 points), Presence Quality (20 points), Brand Recognition (20 points), Share of Voice (10 points), and Market Competition (10 points). Results arrive in under two minutes. No account required.

The $50/month HubSpot AEO product is a continuous monitoring layer. It tracks brand citations across the same three platforms, monitors competitor share of voice, provides prompt-level citation analysis, and generates prioritised recommendations. The entry tier includes 25 tracked prompts.

<h2>Why the Sentiment Weighting Matters</h2>

The most revealing design decision in HubSpot's scoring model is that Sentiment carries 40% of the total score — more than any other dimension. This is not an arbitrary weighting. It reflects a structural reality of AI-mediated discovery that most brands have not yet internalised: AI engines do not just recognise brands, they characterise them. A brand can appear in every AI answer about its category and still lose to a competitor if the AI describes it as 'complex,' 'expensive,' or 'best suited for large enterprises.' The characterisation is the citation.

<h2>What the Research Shows</h2>

HubSpot's State of AEO in 2026 report — based on a survey of 1,100+ global marketers and citation data from Xfunnel's LLM database covering December 2025 through March 2026 — provides the most comprehensive primary research on the category published to date. The headline numbers are significant. AI referral traffic converts at 11.4% compared to 5.3% for organic search, according to Similarweb's September 2025 global ecommerce data. Forty-four percent of marketers have made a business purchase based on a brand discovered through an AI answer. Forty-two percent view AI-recommended brands as more trustworthy.

The content type data is equally actionable. Comparison pages are cited by ChatGPT at a 95% rate — the highest of any content type across the six platforms studied. PR content reaches 92% on ChatGPT. Blog content performs best in Google AI Overviews at 42%. The implication for content strategy is direct: if a brand has limited budget for AEO-specific content creation, comparison pages and PR placements deliver the highest citation return.

<h2>What This Means for the Market</h2>

HubSpot entering AEO is the category's clearest mainstream validation signal to date. The SEO software market followed a similar pattern: a period of startup proliferation (Moz, Majestic, Ahrefs, SEMrush), followed by the entry of established marketing platforms that bundled SEO capabilities into broader suites. HubSpot itself added SEO tools to Marketing Hub in 2017. Eight years later, it is doing the same with AEO.

The implications for brands are straightforward. AEO measurement is no longer an enterprise-only capability. The free Grader removes the cost barrier for a first visibility audit. The $50/month monitoring tier puts continuous citation tracking within reach of any marketing budget. The brands that act on this data now — while most competitors are still in the 'experimental' phase that HubSpot's research identifies — will hold a measurable first-mover advantage as AI search continues to displace traditional search volume.

<h3>References</h3>

[3] Similarweb, Global Ecommerce Traffic Analysis, September 2025.

OpenAI and Yelp's new partnership puts Yelp reviews and ratings inside ChatGPT — but the 'Request a Quote' feature that drives actual leads is reserved for paid Yelp advertisers.

OpenAI and Yelp's new partnership puts Yelp reviews and ratings inside ChatGPT — but the 'Request a Quote' feature that drives actual leads is reserved for paid Yelp advertisers.

On July 23, 2026, Axios reported exclusively that Yelp is licensing its reviews, ratings, photos, and business information to OpenAI. ChatGPT will now surface Yelp content — star ratings, review excerpts, photos, business details — inside responses to local queries. Yelp branding and back-links appear alongside the content, with access to Yelp's 'Request a Quote' feature allowing users to message local service providers without leaving the chatbot interface. The deal is non-exclusive, meaning Yelp can sign similar agreements with other AI companies. Financial terms were not disclosed. Yelp shares surged 8% on the news.

The partnership is a logical extension of Yelp's existing distribution strategy. The company already licenses data to Apple Maps and Amazon's Alexa. CEO Jeremy Stoppelman framed it plainly to Axios: 'If you want to answer local queries, you really need Yelp.' With 330 million cumulative reviews and more than 8 million business listings, Yelp has the depth of structured local data that AI engines need to answer the kinds of questions — 'best plumber near me,' 'top-rated Italian restaurant downtown' — that have historically been Google's domain. As AI-generated noise floods the web, verified human reviews are becoming digital gold. For Yelp, the deal transforms ChatGPT from a potential threat into a massive new distribution engine.

<h2>The Part Nobody Is Talking About</h2>

The Axios story focused on the data licensing deal. What it did not explore in depth is the commercial structure of the feature that actually drives leads: Yelp's Request a Quote.

According to a Yelp sales representative email circulating on LinkedIn, the Request a Quote feature — which lets ChatGPT users contact local service providers for quotes directly inside the chat interface — will appear in local service searches. The catch: the only way a local business is selected for that feature is if they are paying for Yelp Ads.

This is not a minor footnote. It means that while any business with a Yelp listing may appear in organic ChatGPT responses via review data, the action layer — the button that converts a ChatGPT user into an inbound lead — is a paid placement. To generate actual leads from ChatGPT via Yelp, you need to be a Yelp advertiser.

<h2>This Is What Pay-to-Play Looks Like Inside an AI Engine</h2>

The structure mirrors what Google did with Local Services Ads. A business can rank organically in Google Maps and still lose the lead to a competitor who pays for the 'Book Now' or 'Get a Quote' button that sits above the organic results. Yelp is replicating that model inside ChatGPT — organic visibility for reviews, paid placement for conversions.

For local businesses, the implication is direct: AEO is no longer purely an organic discipline. In the local services category, it now has a paid layer. A plumber, contractor, or home services provider who wants to capture leads from ChatGPT queries will need to run Yelp Ads to access the Request a Quote placement. The organic Yelp listing gets them into the answer. The paid placement gets them the call.

This is also the first documented instance of a paid placement requirement inside a major AI answer engine. It will not be the last. As AI platforms license structured data from review networks, directories, and vertical databases, the commercial model that emerges will almost certainly include premium placement tiers — exactly as search did.

<h2>The Yelp-Google History Is a Warning</h2>

Yelp has been through this before. The company previously licensed content to Google before the relationship deteriorated into years of conflict over Google's local search practices. Yelp is now suing Google, alleging the company unfairly favored its own local results over Yelp's. Stoppelman acknowledged the risk to Axios but argued the exposure still benefits the company: 'Ultimately, we believe that if we allow our content outside the walls of just Yelp, and we provide it in useful ways to consumers … value does accrue back to Yelp.'

That argument held with Apple Maps and Alexa, where Yelp's brand remained visible and the platforms did not build competing review products. OpenAI is a different kind of partner. It is building an AI assistant that aspires to replace the need to visit any third-party platform at all. Whether the Yelp-OpenAI relationship follows the Apple Maps model or the Google model is the most important unanswered question in this deal. The closest recent parallel is Reddit's content-licensing deal with OpenAI — a high-visibility arrangement that gave Reddit distribution inside ChatGPT while preserving its brand. Yelp is betting on the same playbook.

<h2>What Local Businesses Should Do Now</h2>

<h2>Market Reaction</h2>

Yelp shares jumped 8% on July 23 when the Axios story broke, closing the session up roughly 5%. The market's read was straightforward: a non-exclusive content deal with the world's most-used AI assistant is a distribution win, not a dilution. Investors who spent years watching Yelp fight Google over local search placement saw the OpenAI deal as the opposite dynamic — a platform actively choosing to embed Yelp rather than build around it.

The Reddit parallel is instructive here too. When Reddit announced its OpenAI content deal in May 2024, shares rose sharply and the company cited AI licensing as a new revenue category in its first earnings report as a public company. Yelp has not disclosed financial terms, but the market is pricing in a similar trajectory: structured, human-generated data commands a premium in an AI landscape increasingly skeptical of synthetic content.

For AEO practitioners, the stock move is a useful signal. It confirms that the market views AI distribution deals as value-additive for data owners — which means more platforms will pursue them. The Yelp-OpenAI deal is unlikely to be the last of its kind in the local and vertical search space.

Three actions are immediately relevant. First, audit your Yelp listing for accuracy — name, address, categories, hours, and photos. The data Yelp feeds to ChatGPT is drawn from your listing. Errors in your listing become errors in ChatGPT's answers about your business. Second, assess whether Yelp Ads make sense for your category. If you are in home services, legal, medical, or any other high-intent local category, the Request a Quote placement in ChatGPT will be a meaningful lead source. The economics of Yelp Ads need to be evaluated against that expanded distribution, not just against Yelp's own platform traffic. Third, monitor how ChatGPT describes your business. The Morning Consult research cited by Search Engine Land found that 65% of Americans have used AI search but only 15% trust it 'a lot.' That trust gap closes as AI answers prove accurate. Businesses that ensure their Yelp data is correct are investing in the accuracy of AI answers about them.

<h3>References</h3>

[3] Morning Consult, AI Search Trust Survey, cited in Search Engine Land, July 2026.

The AEO platform can now design, build, and run marketing agents from inside Claude. The orchestrator-specialist pattern is the new architecture for AI-native marketing stacks.

The AEO platform can now design, build, and run marketing agents from inside Claude. The orchestrator-specialist pattern is the new architecture for AI-native marketing stacks.

On July 20, 2026, Profound became an official Anthropic connector. That is a different thing from a custom MCP. Official connectors are verified by Anthropic and available inside Claude in a couple of clicks, with no config file or setup required. Profound is the first AEO platform to reach that status.

The announcement came in two parts. Part 1, published in May, introduced a knowledge graph and 15 MCP capabilities covering AEO data, prompt tracking, and Profound reporting. The idea was to let Marketing Engineers pull Profound data into Claude without leaving their conversation. Part 2 closes the loop: you can now push actions back into Profound from Claude, not just read from it.

<h2>Three New Capabilities</h2>

The first new capability is prompt generation. You can tell Claude to add ten prompts to a specific category in your Profound account, or to generate variations around a new page, and it calls the right tool, updates your tracked prompt set, and confirms in the same session. No tab-switching.

The second is agent design and execution. Profound Agents are purpose-built marketing specialists: they synthesise AEO data, produce citation-worthy content, or identify content gaps. Previously you built them inside Profound's own interface. Now you describe what you want in plain language inside Claude and the agent is designed and ready to run in the same conversation.

The third is workflow ideation. You can work with the MCP directly in Claude to map out marketing automations before committing to building them. It is a planning layer that sits above the execution layer.

<h2>The Orchestrator-Specialist Pattern</h2>

Profound frames the underlying architecture clearly. Claude is a generalist: broad, capable, good at reasoning across whatever you put in front of it. A Profound Agent is a narrow specialist: it does one marketing job well, on top of your brand's context, Profound's data, and Profound's runtime. Put them together and you get an orchestrator coordinating sub-agents, a mixture of experts.

This is not a novel concept in software engineering. It is, however, a new way to think about how marketing teams will structure AI workflows. The implication is that the value of a platform like Profound is not just the dashboard or the data. It is the runtime that makes agents reusable. An agent built in a Claude conversation runs on the same infrastructure that powers Profound's enterprise customers at scale. What you describe in a sentence becomes a shareable asset your team can deploy repeatedly.

For those who prefer to work inside Profound, the company's own Aim product serves as the orchestrator. For those who live in Claude, Claude is the orchestrator. The MCP is the handoff layer between them.

<h2>What Is Still Missing</h2>

The connector is enterprise-only at launch. Self-serve customers need to request a demo to access it. That is a meaningful constraint: the teams most likely to experiment with agentic AEO workflows are often at smaller, faster-moving companies that are not on enterprise contracts. Profound has signalled that the catalog of available specialists will expand, but the current set covers a focused range of jobs. The roadmap is to widen it.

The broader question is whether the orchestrator-specialist pattern becomes the standard architecture for AI-native marketing stacks, or whether it remains a workflow for a specific type of technically sophisticated marketer. Profound calls this person the Marketing Engineer. The bet is that the category grows as the tools get easier to use.

<h3>References</h3>

A new Profound research report maps the full architecture of ChatGPT Shopping for the first time: how product cards are selected, where commercial data is sourced, and what actually drives rank.

A new Profound research report maps the full architecture of ChatGPT Shopping for the first time: how product cards are selected, where commercial data is sourced, and what actually drives rank.

On July 23, 2026, Profound published the most detailed technical breakdown of ChatGPT Shopping produced to date. The research covers 201,137 unique prompt runs and 812,190 product cards from June 18 to June 25, 2026. The goal was to map the full architecture of the shopping experience from the user's initial query through to the offer list that drives actual clicks to merchant sites.

The report is the third in a series. The first, published in March 2026, tracked roughly two million prompts to establish how often Shopping mode activates. The second reverse-engineered the Shopping trigger itself, finding that category beats intent: naming a shippable product is a six-times lift over baseline, while purchase-oriented language without a product noun barely moves the needle. This third report goes deeper, into the mechanics of how product cards are built once Shopping mode fires.

<h2>The Architecture, Explained</h2>

The key finding is that ChatGPT Shopping operates through two distinct query fanout mechanisms. The first is the standard query fanout that AEO practitioners already know: ChatGPT rephrases the user's prompt into multiple internal sub-queries to build the text narrative. The second, which Profound calls product query fanout, is specific to Shopping mode. ChatGPT also translates the user's prompt into internal product sub-queries used to identify product card candidates. These are separate processes with separate retrieval logic.

Once product card candidates are selected, commercial information for each card is sourced from one of three places. Direct merchant sources account for just under 25% of retrievals. ChatGPT's own internal fallback search index handles a significant share. Google's commercial index covers the remainder. Bing does not appear in the retrieval stack. For the 12.7% of cards sourced from direct product feeds, the logic is simpler: commercial information always comes from the feed itself.

The most operationally important finding is how the offer list connects to the product card. When offers are successfully retrieved for a card, the card's commercial information, including product name, price, and merchant, is always pulled from the rank 1 offer. When no offer is retrieved, ChatGPT falls back to its own product search. Rank 1 in the offer list is not just a conversion lever. It determines what information the product card shows.

<h2>What Drives Product Card Rank</h2>

Profound compared the top-ranked product cards (rank 1) against the bottom-ranked (rank 4 and below) across 415,276 cards. Two factors showed the largest lift. GPT tag presence produced a 144% lift. Median review count produced a 124% lift, with top-ranked cards showing a median of 787 reviews versus 352 for bottom-ranked cards. Promotional pricing produced a modest 13% lift. Offer presence, URL length, and product name length showed no meaningful contribution to rank.

GPT tags are machine-assigned labels based on available product data. They are not set by the merchant directly. Tags in the value and premium clusters are the most prevalent, followed by style, everyday, and classic groupings. The practical implication is that product page and feed data that clearly signals a product's positioning, whether budget, premium, or performance-oriented, increases the probability of a tag being assigned and the card ranking higher.

<h2>The Citation Layer</h2>

Reddit is the most indexed domain in ChatGPT Shopping citations, accounting for roughly one third of all shopping citations. This is consistent with broader findings across non-shopping ChatGPT responses. For brands, it reinforces the case for maintaining a credible presence in Reddit communities relevant to their product categories, not as a direct commerce channel, but as a citation source that influences which products get surfaced and how they are described.

The 87.3% of product cards retrieved from web crawl, rather than product feeds, underscores that organic discoverability still dominates the Shopping surface. Feed integration matters for the 12.7% of cards it covers, and it matters significantly for offer list rank. But the majority of product card selection happens through the same web crawl and search index signals that determine general AI visibility.

<h2>What This Means for Retailers and AEO Practitioners</h2>

Profound draws four operational conclusions from the research. First, product query fanout requires its own optimization strategy. Brands need to align product listings and content with the specific query fanouts ChatGPT uses for product searches, not just the user-facing prompt. Second, ChatGPT fallback search matters: product listings must appear prominently when ChatGPT searches for the exact product name. Third, Google Shopping index signals feed directly into ChatGPT's retrieval stack, so Google Shopping optimization is not separate from AEO for retailers. Fourth, general citation share influences product card candidate selection, meaning content strategy and AEO are not separate workstreams for e-commerce brands.

The research also clarifies the relationship between the offer list and the product card in a way that changes how retailers should think about feed optimization. Getting into the offer list is not just about driving clicks from the sidebar. It determines the commercial information displayed on the card itself. A brand that ranks second in the offer list may still lose the product card display to the rank 1 merchant.

<h3>References</h3>

New 2026 data shows AI has rapidly become a primary product-research channel. But 86% of shoppers verify AI recommendations before buying, and trust still ranks AI sixth among information sources.

New 2026 data shows AI has rapidly become a primary product-research channel. But 86% of shoppers verify AI recommendations before buying, and trust still ranks AI sixth among information sources.

AI has moved surprisingly quickly from experimental shopping tool to mainstream research channel. New data compiled by MarTech from several 2026 studies shows that 43% of U.S. online shoppers have used an AI assistant such as ChatGPT, Gemini, Claude, or Perplexity for product research in the past 90 days. One in five used AI while researching their most recent online purchase worth more than $50.

But the more revealing finding is not simply how many consumers are using AI. It is what they are using it for, and what they do after receiving the answer.

<h2>AI Is Becoming the Starting Point for Shopping Research</h2>

Among consumers already using AI, 46% now begin their purchase research on a standalone AI platform, up from 25% in 2024. Over the same period, the proportion starting with traditional search fell from 43% to 24%. That is a significant behavioural shift in two years.

Consumers are not primarily using AI to discover products. The largest group, 53%, uses AI mainly to compare alternatives and narrow a shortlist. Only 21% primarily use it for final decision-making, while 19% use it primarily for discovery. AI's most established role in shopping today sits squarely in the middle of the purchase journey: compare the options, explain the differences, help me narrow the field.

For brands, that makes appearing in AI answers important. But it makes how the brand is characterized relative to competitors potentially even more important.

<h2>AI Traffic Is Growing Extraordinarily Quickly</h2>

The traffic numbers are striking. Adobe data cited by MarTech shows a 1,200% year-over-year increase in AI-sourced traffic to U.S. retail websites. Those AI-referred shoppers also converted at rates 31% higher than visitors arriving from other sources. Separate L.E.K. Consulting data reports a 176% compound annual growth rate in AI-referred web traffic, while search, social, and direct traffic remained comparatively flat. More than 1.1 billion monthly retail-site visits came from AI in 2025.

The absolute base matters when interpreting enormous percentage growth rates from a relatively young channel. But the direction is difficult to dismiss: AI-generated referrals are becoming a meaningful source of commercial traffic.

<h2>Consumers Are Increasingly Arriving With Their Minds Made Up</h2>

Perhaps the most consequential statistic for marketers is further down the funnel. Thirty-one percent of AI users say their purchase decision is largely made before they reach the brand or retailer's website, compared with 26% two years ago. At the same time, average website visit duration declined 6% across the 13 categories studied between March 2024 and March 2026. Household essentials fell 18%, while healthcare services and beauty and personal care each declined 16%.

That does not establish that AI caused the decline. But taken together, the findings point toward an important change in the role of the website. Consumers who have already asked AI to compare products, summarize reviews, and explain differences may arrive at a brand site with considerably more of their evaluation already completed. The website may increasingly be where an AI-informed consumer validates a decision already taking shape, not where consideration begins.

<h2>But AI Still Has a Trust Problem</h2>

This is where the data gets particularly interesting. Despite rapid adoption, 86% of U.S. shoppers who used AI for product research verified its recommendation using another source before purchasing. And AI still ranks only sixth among consumers' most trusted sources of product information. Family and friends rank first, followed by online customer reviews, expert publications or professionals, brand and retailer websites, and Reddit or other online forums. AI ranks ahead only of YouTube creators and reviewers.

So consumers are clearly using AI. They just are not necessarily trusting it enough to stop there.

<h2>Price Changes How Much Consumers Trust AI</h2>

One of the clearest findings involves purchase value. For purchases under $50, 42% of shoppers said they would trust an AI recommendation without checking another source. For purchases above $500, just 5% would. That suggests AI's role in shopping cannot be understood with one universal model. The same consumer who is comfortable accepting an AI recommendation for an inexpensive household product may want extensive verification before acting on its recommendation for a laptop, vacation, or financial product.

Interestingly, that is not preventing consumers from using AI in higher-consideration categories. Reported AI research usage reaches 71% for travel, 65% for consumer electronics, and 62% for financial products. Every category tracked in the L.E.K. data was at 47% or higher. Consumers appear willing to use AI extensively for important decisions without necessarily being willing to surrender the decision to it.

<h2>Autonomous Shopping Is Still Further Away</h2>

The same pattern appears when consumers are asked about AI agents. According to the Adobe data cited by MarTech, 47% of customers would be comfortable having their personal AI agent interact with a brand's human representative. Consumers were less comfortable allowing agents to exchange personal information or authorize purchases. Gartner similarly found that consumers are much more receptive to AI finding coupons and deals than actually completing transactions for them.

NielsenIQ forecasts that agentic shoppers could eventually represent $190 billion to $385 billion in U.S. ecommerce spending by 2030, roughly 10% to 20% of online retail sales. Today's consumer behaviour suggests there is still considerable distance between 'help me shop' and 'shop for me.'

<h2>What the Data Means for AEO</h2>

The immediate AEO implication is more concrete than the grand predictions about autonomous agents. AI is already being used at meaningful scale to research products, compare alternatives, create shortlists, and shape decisions before consumers reach brand websites. That means measuring whether a brand appears in AI answers is increasingly important.

But the 53% comparison-and-shortlisting figure suggests marketers should be equally concerned with the next question: when AI compares your brand with competitors, what does it say, and do you survive the shortlist?

The trust findings add another dimension. With 86% of AI shoppers still checking recommendations elsewhere, the sources consumers use for verification, including reviews, expert publications, brand websites, and online communities, remain critically important. AI shopping is not replacing the broader information ecosystem. For now, it appears to be reorganizing it.

"AI shopping has gone mainstream. AI shopping without human verification hasn't."

<h3>References</h3>

[2] Adobe Analytics, AI-driven traffic to U.S. retail websites, 2026.

[3] L.E.K. Consulting, AI-referred web traffic compound annual growth rate analysis, 2025.

[4] NielsenIQ, Agentic commerce forecast, 2025.

HubSpot's free AEO scoring tool is not a web crawl. It is a structured LLM self-assessment. Here is the full architecture, the scoring weights, and what the results actually tell you.

HubSpot's free AEO scoring tool is not a web crawl. It is a structured LLM self-assessment. Here is the full architecture, the scoring weights, and what the results actually tell you.

When HubSpot launched its free AEO Grader in July 2026, the marketing positioned it as a tool that 'reveals what AI engines are saying about your brand.' That framing is accurate but incomplete. To understand what the scores actually mean, you need to know how the tool works under the hood. So we reverse-engineered it.

The short version: the HubSpot AEO Grader is not a web crawl. It does not simulate a real user query and observe what ChatGPT, Perplexity, or Gemini returns. It sends your brand information directly to each AI engine via API and asks the model to evaluate itself. The score you receive is the LLM's own assessment of how well it knows your brand. That is a meaningful distinction.

<h2>The Architecture</h2>

The grader takes four inputs: company name, geography, products or services, and industry. Once submitted, it fires three parallel API calls, one to each engine: OpenAI (specifically GPT-5.4 mini, not GPT-5 or GPT-4o), Perplexity, and Gemini. Each call has a 90-second timeout. The request payload for each call includes the four input fields plus a session hash, language code, and tracking source identifier.

Each engine call hits a scoring endpoint at <code>aims-ai-grader.hubwt.com/v2/production/score</code>, with a queue-based fallback at <code>wtcfns.hubspot.com/wt-ai-grader-api/v2/score</code>. A separate call to <code>/v2/llm_reports</code> retrieves the written interpretation of the scores. The PDF report is generated via a third service at <code>pdf.hubwt.com</code>. The full written report, brand archetype, confidence level, and improvement recommendations are gated behind a form submission.

HubSpot describes the scoring as 'deterministic' and uses structured output with JSON schema enforcement and retry logic to get consistent score formats from each LLM. The scoring rubric is applied by the model itself, not by a deterministic algorithm on HubSpot's servers. The model receives the brand context and the scoring criteria and returns a structured JSON object with numeric scores.

<h2>The Scoring Framework</h2>

The total score is 100 points distributed across five dimensions. Sentiment carries 40 points, the largest single weight. Presence Quality and Brand Recognition each carry 20 points. Share of Voice and Market Competition each carry 10 points.

Sentiment (40 points) covers three sub-dimensions: general sentiment (the overall positive, negative, or neutral tone of AI descriptions), contextual sentiment (how tone varies across topics such as customer support versus product innovation), and source-based sentiment (the credibility of sources influencing the AI's characterization). The 40% weighting reflects a structural reality: AI engines do not just recognize brands, they characterize them. A brand that appears in every AI answer about its category but is consistently described as 'complex' or 'best suited for large enterprises' will lose to a competitor described as 'intuitive' and 'fast to deploy.'

Presence Quality (20 points) measures mention depth, source quality, and data richness. Brand Recognition (20 points) measures how widely and specifically the AI can discuss the brand beyond surface acknowledgment. Share of Voice (10 points) measures the brand's rank relative to competitors in AI-generated responses. Market Competition (10 points) assesses how AI positions the brand relative to category peers.

<h2>What the Scores Actually Tell You</h2>

To calibrate the tool, we ran it on Salesforce. The results across the three engines were: OpenAI 87/100, Perplexity 85/100, Gemini 83/100. Brand Recognition was near-perfect across all three (19/20). Market Competition was perfect across all three (10/10). The variation appeared in Sentiment, where Gemini scored Salesforce 30/40 versus OpenAI's 33/40 and Perplexity's 34/40, and in Share of Voice, where Perplexity returned 5/10 versus 8/10 from both OpenAI and Gemini.

The Salesforce result illustrates the tool's primary use case: identifying which engine is characterizing your brand less favorably and in which dimension. A Sentiment gap between Gemini and OpenAI is actionable. It tells you that the sources shaping Gemini's understanding of your brand are producing a different characterization than those shaping OpenAI's. The next question, which the grader does not answer directly, is which sources those are.

<h2>Three Limitations Worth Understanding</h2>

The first limitation is the model choice. The grader uses GPT-5.4 mini, not GPT-5 or GPT-4o. This is a cost optimization. For well-known brands like Salesforce, the mini model has sufficient training data to produce reliable scores. For smaller or newer brands, the mini model may have less complete knowledge than the full model, which could produce scores that understate actual visibility in ChatGPT's consumer-facing product.

The second limitation is the meta-evaluation problem. The grader does not ask the internet what it thinks of your brand. It asks the LLM what the LLM thinks of your brand. The score reflects the model's self-reported confidence in its brand knowledge, filtered through a scoring rubric the model applies to itself. That is not the same as observing what the model actually returns when a real user asks a real question about your category.

The third limitation is the snapshot problem. The free grader produces a single point-in-time assessment. AI model knowledge changes as models are updated and fine-tuned. A score from July 2026 may not reflect the same model state as a score from October 2026. This is the primary commercial argument for HubSpot AEO, the $50/month monitoring product that tracks changes over time.

<h2>What It Is Good For</h2>

Despite those limitations, the grader is a genuinely useful diagnostic for three purposes. First, it provides a fast cross-engine comparison. Seeing that Gemini scores your brand 12 points lower than OpenAI on Sentiment is a signal worth investigating, even if the precise number is not empirically grounded. Second, it surfaces the dimension breakdown. Knowing that your Brand Recognition is strong but your Share of Voice is weak tells you something different than knowing your overall score is 72. Third, it is free and requires no account, which removes the cost barrier for a first AEO audit.

The tool is best understood as a structured prompt to three LLMs, not a measurement instrument. The output is the LLM's self-assessment of your brand's presence in its training data, organized into a consistent scoring framework. Used with that understanding, it is a reasonable starting point for an AEO audit. Used as a definitive measurement of AI visibility, it will mislead you.

<h3>References</h3>

A new MediaPost commentary challenges one of the biggest assumptions in AEO: that more AI visibility automatically creates more business value.

A new MediaPost commentary challenges one of the biggest assumptions in AEO: that more AI visibility automatically creates more business value.

As Answer Engine Optimization continues to gain momentum, most conversations still revolve around one question: how visible is my brand in AI? A new opinion piece published today in MediaPost argues that marketers may be asking the wrong question.

The article does not dismiss AI search. It challenges the industry's growing tendency to treat AI visibility as a success metric without first proving that it actually influences consumer behaviour. It is an important conversation, and one AEO practitioners should welcome.

<h2>Visibility Is a Metric. Value Is the Objective.</h2>

Most AEO platforms today measure some combination of AI visibility, share of voice, citation frequency, brand mentions, and AI referrals. These are useful indicators. But they are also intermediate metrics.

The MediaPost article argues that marketers still struggle to answer a more fundamental question: does increasing AI visibility actually change commercial outcomes? A brand appearing in 40% of AI responses instead of 20% is interesting. Whether that translates into greater consideration, more branded search, increased website visits, or higher sales is a different question entirely.

<h2>Attribution Remains the Industry's Biggest Challenge</h2>

One reason this debate is emerging now is that AI search does not behave like traditional search. Users often receive synthesized answers without clicking through to a website. That means many familiar SEO metrics — traffic, rankings, and clicks — become less informative. The industry has responded by measuring AI visibility instead. But visibility alone does not establish commercial impact. It tells us a brand appeared. It does not tell us whether that appearance influenced the decision. That distinction matters as organizations begin allocating meaningful AEO budgets.

<h2>The Next Generation of Measurement Needs to Go Deeper</h2>

Rather than asking only whether a brand appeared, marketers may increasingly need to ask: Was the brand recommended? How was it characterized? Which competitors appeared alongside it? Which claims were used to describe it? Did the recommendation influence subsequent consumer behaviour? Those questions move beyond visibility toward understanding why AI recommends certain brands, and whether those recommendations actually matter.

<h2>A Healthy Debate for a Young Industry</h2>

Every emerging discipline experiences a period where measurement evolves faster than proof. SEO went through it. Social media did. Influencer marketing did. AEO is no different. Today's visibility metrics are valuable because they provide the first systematic view into how brands appear inside AI-generated answers. But as the industry matures, marketers will increasingly expect evidence that links those metrics to business outcomes.

That does not diminish the importance of AI visibility. It raises the bar for what comes next.

<h2>The Bottom Line</h2>

The MediaPost article asks a question the AEO industry should take seriously: if AI visibility is becoming a marketing KPI, how do we know it creates value? The answer is unlikely to be abandoning visibility measurement. It is building the next layer on top of it — connecting visibility to recommendation, recommendation to consumer behaviour, and ultimately consumer behaviour to commercial performance. For an industry still in its early stages, that is exactly the kind of debate worth having.

<h3>References</h3>

Google's latest AI search change may have turned AI participation from a strategic choice into a competitive necessity.

Google's latest AI search change may have turned AI participation from a strategic choice into a competitive necessity.

For much of the past year, publishers have debated whether to allow their content to be used in Google's AI-powered search experiences. Some viewed it as a licensing question. Others viewed it as a traffic question. Google's latest search experiment suggests it may increasingly become a visibility question.

According to reporting from Search Engine Journal, Google is now placing Top Stories carousels directly inside some AI Overviews, rather than displaying them as a separate search feature further down the page. NewzDash, which tracks publisher visibility, estimates this is already occurring in roughly 15.5% of U.S. searches where Top Stories appear. At first glance, this might appear to be a relatively minor interface change. It is not.

<h2>AI Is No Longer a Separate Destination</h2>

For the past two years, marketers have often discussed AI search as though it existed alongside traditional search. That distinction is becoming harder to defend. Google is not simply adding AI to Search. It is increasingly embedding traditional search features inside AI-generated experiences.

The implication is significant. Publishers who choose to opt out of AI may eventually find themselves opting out of search experiences they previously expected to participate in. Whether that becomes Google's long-term approach remains to be seen. But the strategic calculation has clearly changed.

<h2>Participation Is Becoming the Default</h2>

Many publishers initially hoped they could preserve traditional search visibility while declining participation in AI-generated answers. That separation may become increasingly difficult to maintain. As AI features become more deeply integrated into Google's core search experience, participating in AI may no longer be a question of chasing a new channel. It may simply become part of participating in Search itself.

For marketers, that has broader implications than publisher strategy. If Google continues integrating AI-generated answers with established discovery features, brands may find that AI optimization shifts from an emerging capability to a baseline expectation.

<h2>AEO Is Becoming Less Optional</h2>

One of the defining characteristics of new marketing channels is that early adoption is usually voluntary. Eventually, the channel becomes infrastructure. We are beginning to see that transition with AI search. Whether or not Google's current experiment becomes permanent, the direction is becoming clearer. AI is moving from an additional search feature to a foundational layer within search itself.

For brands, the question is becoming less 'Should we optimize for AI?' and more 'Can we afford not to?'

<h3>References</h3>

[2] NewzDash, Publisher Visibility Tracking, July 2026.

A new enterprise survey suggests AEO has officially moved from experimentation to budget line item. But the more interesting story is not the percentage — it is what enterprises are actually investing in.

A new enterprise survey suggests AEO has officially moved from experimentation to budget line item. But the more interesting story is not the percentage — it is what enterprises are actually investing in.

Every emerging marketing discipline eventually reaches the same milestone: someone starts putting real money behind it. According to the 2026 CMO Investment Report from Conductor, 94% of enterprise organizations plan to increase their investment in AEO and GEO during 2026. The report surveyed more than 250 enterprise digital leaders. On its face, that is an impressive statistic. But the more interesting story is not the percentage. It is what enterprises are actually investing in.

<h2>This Is No Longer an SEO Side Project</h2>

The report suggests that organizations are moving beyond experimentation. Among respondents, 93% are building AEO capabilities internally, rather than relying primarily on external agencies. That is an important signal. When companies begin hiring people, developing internal processes, and integrating new measurement frameworks, a capability has typically crossed the line from curiosity to operational priority. Whether every organization succeeds is another question. But the budget conversation has clearly begun.

<h2>Success Metrics Are Changing</h2>

Perhaps the most encouraging finding is not the investment itself. It is how organizations say they intend to measure it. According to the report, leading organizations are shifting away from traditional SEO metrics like traffic and rankings toward measures such as conversions, brand sentiment, AI search market share, and broader business outcomes. That evolution matters. AEO should not simply become another race for impressions. Marketers ultimately need to understand whether AI visibility changes consumer behaviour and business performance. The industry still has work to do in proving those relationships, but it is encouraging to see enterprises already thinking beyond rankings alone.

<h2>Original Information Is Becoming a Competitive Advantage</h2>

Another theme running through the report is the growing importance of first-party research, structured data, and original content. High-maturity organizations appear to be investing more heavily in proprietary information than those at earlier stages of AEO adoption. That trend aligns with a broader pattern visible across AI search. Large language models do not simply retrieve webpages. They synthesize information from across the web. Brands that contribute genuinely original data, research, and evidence are more likely to become part of that information ecosystem than brands that simply restate what already exists.

<h2>A Note of Caution</h2>

It is important to interpret these findings in context. This is a vendor-sponsored survey conducted among enterprise digital leaders — organizations already much more likely than the average business to be actively evaluating AI search. The results should not be automatically generalized to every company or every market. Even so, the report captures something important. Enterprise marketing teams are no longer asking whether AI search matters. They are asking how quickly they need to respond.

<h2>The Bigger Takeaway</h2>

The headline is not really '94%.' The headline is that AEO has reached a new stage of maturity. Budgets are forming. Teams are being built. New success metrics are emerging. And organizations increasingly see AI search as something requiring dedicated strategy rather than occasional experimentation. Whether today's investments ultimately produce the expected returns remains an open question. But one thing appears increasingly clear: AEO has entered the enterprise planning cycle. That alone is a meaningful milestone for the industry.

<h3>References</h3>

A new paper from Alibaba researchers on e-commerce intent generation has unintentionally exposed a fundamental flaw in how the AEO industry measures brand visibility.

A new paper from Alibaba researchers on e-commerce intent generation has unintentionally exposed a fundamental flaw in how the AEO industry measures brand visibility.

A newly published research paper from Alibaba researchers may have unintentionally highlighted one of the biggest challenges facing brands trying to measure performance in AI search. The paper, <em>Improving Item Discoverability in e-Commerce Search via Related Intent Generation</em>, is not about Answer Engine Optimization. It is focused on improving product discovery in AI-powered shopping systems. But hidden within its proposed architecture is an insight with significant implications for how brands evaluate AI visibility.

<h2>AI Is Beginning to Search Across Intent Networks</h2>

For decades, search engines treated each query as an isolated request. A user searched for 'best running shoes' and the system attempted to retrieve the best answer for exactly that query. The Alibaba researchers propose something fundamentally different. Instead of searching immediately, their system first generates a collection of related consumer intents before retrieving products. A simple search for 'best running shoes' might expand internally into concepts such as marathon running, trail running, stability shoes, lightweight shoes, wide feet, injury prevention, beginner runners, and budget options. Rather than treating these as separate searches, the system recognises them as part of the same underlying decision space. In other words, AI begins reasoning across a network of related intents instead of responding to a single prompt.

<h2>Why This Matters for AI Search Measurement</h2>

Today's AI visibility tools generally evaluate brands using a relatively small number of prompts — for example, 'best running shoes,' 'top running shoes,' 'best shoes for runners' — and calculate visibility based on whether a brand appears within those responses. But if modern AI systems are increasingly expanding a user's intent internally, a fundamental question emerges: are we measuring the right thing? If AI is effectively evaluating brands across dozens of related consumer intents, then a handful of manually selected prompts may only capture a small portion of the decision landscape. A brand may perform exceptionally well for general running shoes while disappearing entirely once the conversation shifts toward stability, injury prevention, or marathon performance. Those differences remain largely invisible when measurement is limited to isolated prompts.

<h2>From Individual Prompts to Prompt Domains</h2>

The research suggests that marketers may need to rethink the basic unit of measurement. Instead of asking 'How did my brand perform for this prompt?', a more useful question may become 'How does my brand perform across the entire family of related intents surrounding this consumer decision?' This shifts the focus from measuring individual prompts toward measuring what might be described as a Prompt Domain — a collection of semantically related prompts representing the broader decision space associated with a consumer need. As AI systems become better at understanding intent rather than simply matching keywords, evaluating entire prompt domains may become increasingly representative of how brands are actually surfaced.

<h2>Questions the Industry Should Be Asking</h2>

Although this research focuses on e-commerce retrieval, its implications extend well beyond shopping. The study provides additional evidence that AI search is becoming increasingly intent-driven rather than query-driven. That raises an important challenge for the AEO industry. Should AI visibility be measured across entire Prompt Domains rather than individual prompts? How consistently do brands perform as consumer intent shifts? Which types of intent cause brands to disappear from AI recommendations? What new metrics will be needed as AI search moves beyond keyword matching? The research does not answer those questions. But it strongly suggests they are the right questions to begin asking.

<h3>References</h3>

[1] Alibaba Research, <em>Improving Item Discoverability in e-Commerce Search via Related Intent Generation</em>, 2026.

A Tom's Guide reader survey reveals that consumers are using AI as a decision assistant, not a decision maker — and that distinction changes everything about how brands should measure AI search performance.

A Tom's Guide reader survey reveals that consumers are using AI as a decision assistant, not a decision maker — and that distinction changes everything about how brands should measure AI search performance.

For the past two years, the conversation around AI search has been dominated by one question: 'How do I get ChatGPT to recommend my brand?' A recent reader discussion published by Tom's Guide suggests that may be the wrong question entirely. The article collected responses from readers about how they use AI search, and one theme emerged repeatedly: people are increasingly turning to AI because traditional search engines have become cluttered with ads, SEO content, and low-quality results. Yet many of those same users also expressed discomfort with AI making decisions on their behalf. They do not want AI to replace judgment. They want AI to make judgment easier. That distinction matters.

<h2>AI Is Becoming the First Step, Not the Final Step</h2>

Many marketers still imagine AI search as a replacement for Google: a user asks a question, the AI recommends a brand, the customer buys. Reality appears to be more nuanced. The emerging decision journey looks more like this: AI provides a recommendation, the user asks why, the user evaluates the evidence, the user validates the recommendation, and only then does the decision happen. AI is increasingly acting as a decision assistant, not a decision maker.

<h2>The New Question Is Not 'Did AI Recommend Me?'</h2>

If users continue to seek explanations rather than blind recommendations, brands need to rethink how they measure success. Simply appearing in an AI response will not be enough. Brands will increasingly need to understand why they were recommended, what claims the AI relied on, what evidence supports those claims, and whether those claims would withstand customer scrutiny. Those questions are fundamentally different from traditional SEO metrics. They are also much more valuable.

<h2>Visibility Is Only the Beginning</h2>

The AI optimization industry has understandably focused on visibility: did your brand appear, how often, and where? Those remain important questions. But visibility is only the first step in a much larger decision process. The next questions are far more interesting. What does AI believe about your brand? Why does it believe those things? Which pieces of evidence reinforce those beliefs? Which narratives are influencing customer decisions? Those are measurement problems, not search problems.

<h2>The Shift From Search to Decision Intelligence</h2>

For years, search optimization has been about increasing discoverability. AI introduces a different challenge. Brands now need to understand how they are represented during customer decision-making. That is a richer problem. It combines visibility with credibility, evidence, associations, and the claims that ultimately shape trust. The brands that succeed will not simply be the ones that appear most often. They will be the ones whose strongest claims are consistently supported by credible evidence and reinforced across AI systems.

<h2>AEO Updates Takeaway</h2>

The future of AI optimization is unlikely to be won by the brands that simply maximise visibility. It will be won by the brands that understand — and actively shape — how AI constructs, explains, and reinforces their reputation during real customer decision journeys. As AI becomes a standard part of consumer decision-making, the industry will need to move beyond simple visibility metrics toward understanding how AI represents brands when customers are actually making decisions. That may prove to be the metric that matters most.

<h3>References</h3>

The companies that recognise the difference between traffic and brand distribution will be the ones that win in AI search.

The companies that recognise the difference between traffic and brand distribution will be the ones that win in AI search.

For more than twenty years, digital marketing operated on a simple model: create content, rank in Google, earn the click, convert the visitor. That model is disappearing. Publishers are reporting significant declines in search traffic as Google increasingly answers questions directly with AI-generated responses instead of sending users to websites. At the same time, AI assistants like ChatGPT, Claude, Gemini, and Perplexity are becoming destinations in their own right. Most discussions about this change focus on traffic. That is understandable — but it is also shortsighted. The more important change is that AI is becoming a new distribution layer for brands.

<h2>The Click Is No Longer the Product</h2>

Historically, Google acted as a directory. It pointed people toward websites. Today, AI increasingly acts as an interpreter. Instead of returning ten links, it returns an answer. Your brand may be part of that answer — or it may not. Either way, the customer often receives what they need without ever visiting your website. Publishers are increasingly adapting their content strategies for AI systems rather than relying solely on referral traffic, because the economics of search are changing.

<h2>A Fundamental Shift in Thinking</h2>

Many organisations are still asking: 'How do we get more AI traffic?' That may soon become the wrong question. A better question is: 'How is AI representing our brand?' Those are very different problems. Traffic measures visits. Representation measures influence. A customer who never visits your website may still decide to buy from you because an AI system recommended your company, repeated your key claims, and supported those claims with credible evidence. That is brand distribution — not website traffic.

<h2>AI Is Becoming a Brand Distribution Platform</h2>

Consider how a prospective customer now researches a purchase. Instead of opening ten browser tabs, they ask: 'Who are the best executive coaching firms?' or 'Which CRM should a mid-sized software company choose?' The AI synthesises thousands of documents into a single recommendation. In that moment, AI is not behaving like a search engine. It is behaving like a media channel — selecting which brands deserve attention and which narratives deserve repetition.

<h2>Why This Matters</h2>

Traditional SEO optimised webpages. AI optimisation increasingly influences perceptions. The questions brands need to answer are changing. Instead of asking 'Did we rank?', they are asking: Were we recommended? Which claims did AI repeat? What evidence supported those claims? Which competitors were positioned beside us? Those are fundamentally different measurement problems — and they require fundamentally different tools.

<h2>The New Scorecard</h2>

For years, marketers measured success through impressions, clicks, and conversions. In an AI-first world, those metrics remain important — but they are no longer sufficient. Brands also need to understand how they are represented inside AI systems during real customer decision-making. The companies that adapt first will not simply generate more traffic. They will become the brands AI consistently recommends, explains, and reinforces. And in the next era of digital marketing, that may be the distribution channel that matters most.

<h2>AEO Updates Takeaway</h2>

The biggest disruption is not that AI is reducing website traffic. It is that AI is becoming a new layer of brand distribution. The organisations that continue measuring only clicks may miss the more important question: how is AI representing your brand when customers are making decisions?

What a new chemistry research platform called AskChem reveals about the direction AI search is heading — and why the claim, not the document, may become the primary unit of brand representation.

What a new chemistry research platform called AskChem reveals about the direction AI search is heading — and why the claim, not the document, may become the primary unit of brand representation.

One of the most interesting AI papers published this week is not about ChatGPT, Google, or search engines. It is about chemistry. Researchers behind AskChem argue that we have been retrieving the wrong thing for decades. Instead of searching for papers, they propose searching for claims. Every paper is decomposed into small, verifiable, evidence-backed claims, each linked directly to its source, then organised into taxonomies and evidence graphs for retrieval and synthesis. At first glance, that sounds like a niche problem for scientists. It is not. It may represent the direction AI search is heading.

<h2>Documents Were Built for Humans</h2>

Traditional search engines were designed around documents. You searched for information, the search engine returned webpages, and you synthesised the answer yourself. Large language models changed that. They no longer retrieve documents for people — they synthesise information from many sources into a single response. That means the document is no longer the most useful unit of knowledge. The claim is.

<h2>Claims Are the Building Blocks of AI</h2>

When ChatGPT answers a question, it does not simply repeat webpages. It constructs a response from hundreds of smaller ideas. Consider the question: 'Why is Volvo considered one of the safest car manufacturers?' The answer is not one document. It is a collection of claims: Volvo invented the three-point seatbelt; Volvo consistently performs well in independent crash testing; Volvo has invested heavily in automotive safety research; Volvo's vehicles receive high safety ratings from independent organisations. Each of those is an individual claim. Together, they create the broader association: Volvo equals safety. AskChem treats those claims as the primary object of retrieval rather than the documents that contain them.

<h2>This Has Implications Far Beyond Science</h2>

Marketing has traditionally focused on keywords. SEO shifted attention toward webpages. AI optimisation may shift attention again — toward claims. Brands do not really want AI to remember every page they have published. They want AI to consistently remember the right facts. Those facts are claims. AskChem indexes 2.4 million structured claims from 147,000 papers, links each claim to source evidence, organises them into multiple taxonomies, and even identifies contradictions across the literature. As AI search evolves, brands may need to think the same way.

<h2>The Future May Be Claim-Centric</h2>

This raises a different question for marketers. Instead of asking 'How many times did AI mention my brand?', they may soon ask: Which claims does AI consistently associate with my brand? Which claims are supported by credible evidence? Which claims are influencing recommendations? Which important claims are missing entirely? Those questions move beyond visibility. They begin measuring how AI constructs an understanding of a brand — a network of reinforcing evidence rather than a single marketing message.

<h2>AEO Updates Takeaway</h2>

The most valuable unit of knowledge in AI search may no longer be the document. It may be the claim. The brands that understand, strengthen, and support their most influential claims could have a significant advantage as AI systems become increasingly claim-centric rather than document-centric.

<h3>References</h3>

[1] AskChem Research Team, <em>AskChem: Claim-Centric Retrieval for Scientific Literature</em>, 2026.

Why the biggest AI story this week isn't about OpenAI losing — it's about the end of a one-model world, and what that means for every brand trying to measure AI visibility.

Why the biggest AI story this week isn't about OpenAI losing — it's about the end of a one-model world, and what that means for every brand trying to measure AI visibility.

For much of the last three years, the AI conversation has revolved around one company. OpenAI defined the market. ChatGPT introduced millions of people to generative AI. Developers built around its APIs. Businesses assumed OpenAI would remain the default platform for enterprise AI. That assumption is becoming increasingly difficult to defend. A new Wall Street Journal analysis argues that Anthropic has overtaken OpenAI on several key business metrics, particularly in enterprise adoption. Claude Code has become the preferred coding platform for many software teams, while Anthropic has surpassed OpenAI in both revenue and private valuation. OpenAI is responding aggressively with new models, product restructuring, and broader platform ambitions — but the competitive landscape has fundamentally changed. Most people will read this story as a battle between two companies. They are missing the bigger picture.

<h2>There Is No Longer 'The AI'</h2>

For the first time since ChatGPT launched, it is becoming clear that there is no single AI ecosystem. Different models are beginning to specialise. Some excel at coding. Others at reasoning. Others at research. Others at enterprise workflows. This is exactly what happens when a technology matures. The market fragments.

<h2>Why This Matters for Brands</h2>

Many organisations are still approaching AI optimisation as if there is one destination to optimise for. 'How do we rank in ChatGPT?' That question is becoming obsolete. Increasingly, brands need to ask: How does ChatGPT represent our brand? How does Claude represent our brand? How does Gemini represent our brand? Where do those representations differ, and why? The answers may not be the same.

<h2>AI Models Are Developing Their Own Brand Memories</h2>

Every frontier model has different training data, retrieval strategies, reasoning behaviour, safety policies, and citation preferences. As a result, they can develop different understandings of the same company. One model may consistently associate a brand with innovation. Another may emphasise price. A third may not recommend the brand at all. Those differences are no longer edge cases. They are becoming an expected part of the AI ecosystem.

<h2>Measurement Becomes More Important Than Optimisation</h2>

As the number of influential AI models grows, optimising for a single platform becomes less valuable. Measurement becomes more valuable. Brands need to understand: Where are they recommended? Where are they absent? Which claims are consistently repeated? Which sources influence each model? How does their representation differ across AI systems? Those questions cannot be answered by monitoring one chatbot. They require a model-agnostic measurement framework.

<h2>AEO Updates Takeaway</h2>

The most important takeaway from OpenAI's changing position is not that Anthropic is winning today. It is that the AI market is fragmenting. As multiple models become influential, brands will need to stop thinking about optimising for ChatGPT and start measuring how they are represented across an increasingly diverse AI ecosystem. That is a fundamentally different challenge from traditional SEO — and it is likely to become one of the defining measurement problems of the AI era.

<h3>References</h3>

[1] Wall Street Journal, <em>Anthropic Has Surpassed OpenAI on Key Business Metrics</em>, July 2026.

Discovered Labs' seven-principle framework is one of the most structured attempts yet to formalise what AI-ready content looks like. It is also a sign that the discipline is growing up.

Discovered Labs' seven-principle framework is one of the most structured attempts yet to formalise what AI-ready content looks like. It is also a sign that the discipline is growing up.

One of the clearest signs that AI optimisation is maturing is that companies are beginning to publish structured methodologies instead of relying on vague advice about writing for AI. Discovered Labs' CITABLE Framework is one of the strongest examples to date. Rather than treating AI optimisation as a collection of SEO tactics, CITABLE presents a structured methodology for engineering content that AI answer engines — including ChatGPT, Claude, Perplexity, and Google AI Overviews — can more reliably retrieve, verify, and cite. Whether or not every element ultimately proves to be predictive of AI citations, the framework represents an important step forward because it encourages organisations to think systematically about how AI systems consume information.

<h2>Understanding the CITABLE Framework</h2>

CITABLE is built around seven design principles intended to improve AI readability and citation potential. The first, Clear Entity and Structure, asks content teams to identify the primary entity and communicate the core answer immediately so AI systems understand exactly what the content is about. The second, Intent Architecture, organises content around the user's primary question while anticipating logical follow-up questions. Third-Party Validation requires that important claims be supported by independent evidence from credible external sources. Answer Grounding ensures key statements are specific, verifiable, and supported by evidence rather than broad marketing language. Block-Structured for RAG asks that content be divided into logical sections that retrieval systems can easily extract and reuse. Latest and Consistent addresses the need to maintain current information while ensuring consistency across all published content. Finally, Entity Graph and Schema calls for the use of structured data and clearly defined entities to help AI systems understand relationships between concepts. Taken together, these principles represent one of the clearest attempts yet to formalise what AI-ready content looks like.

<h2>A Sign That the Industry Is Growing Up</h2>

Perhaps the most interesting aspect of CITABLE is not any individual principle but what the framework represents. For years, search optimisation largely revolved around keywords, rankings, and technical SEO. AI optimisation is beginning to look different. The conversation is shifting toward knowledge structure, evidence, entity clarity, verifiable claims, answer quality, and information architecture. In other words, the discipline is becoming less about optimising pages and more about organising knowledge. That is an important evolution, and it is one that frameworks like CITABLE are actively accelerating.

<h2>No Single Framework Will Define AI Optimisation</h2>

As with traditional SEO, it is unlikely that AI optimisation will ever be defined by a single methodology. Different frameworks will almost certainly emerge to address different aspects of the problem: content engineering, technical implementation, measurement, experimentation, governance, analytics, and workflow integration. That diversity should be viewed as a positive development. Healthy industries rarely converge around one model. They evolve through multiple complementary approaches, each solving a different part of a complex problem.

<h2>Where CITABLE Fits</h2>

Viewed through that lens, CITABLE occupies an important role. It provides organisations with a practical way to think about how content should be structured so AI systems can more easily retrieve, understand, and reference it. For content teams, agencies, and marketers producing AI-ready content, that represents a useful contribution. It does not claim to answer every question surrounding AI optimisation, and it does not need to. Instead, it helps advance one important part of a rapidly developing discipline.

<h2>AEO Updates Takeaway</h2>

The emergence of structured methodologies like CITABLE is a positive sign for the AI optimisation industry. Rather than relying on generic best practices, organisations are beginning to think more systematically about how AI systems retrieve and understand information. As the category matures, a growing ecosystem of complementary frameworks is likely to develop, each addressing different aspects of AI optimisation. If that happens, the discipline will become stronger, more measurable, and ultimately more valuable for brands.

<h3>References</h3>

[1] Discovered Labs, <em>The CITABLE Framework for AI-Ready Content</em>, 2026.

A new study of 2,000 US AI users reveals that platform choice is organised by task, trust is earned not inherited, and the privacy paradox is really an information problem.

A new study of 2,000 US AI users reveals that platform choice is organised by task, trust is earned not inherited, and the privacy paradox is really an information problem.

A new working paper from Profound researcher Jennifer Zou offers the most granular picture yet of how US adults actually use AI assistants — which platform they choose, what they use it for, how much they trust it, and what they would pay to control their data. The study surveyed 1,999 US adults in June 2026, weighted to the AI-using population rather than the general adult population, and its findings challenge several assumptions that underpin how brands and marketers think about the AI assistant landscape.

<h2>The market is concentrated but not uniform</h2>

ChatGPT is the primary assistant for 58% of users and Gemini for 25%, making the two platforms dominant in aggregate. But the aggregate view obscures a more interesting reality at the task level. Claude holds 33% primary share within coding tasks — against just 7% overall — nearly matching ChatGPT (39%) and far exceeding Gemini (15%). Copilot roughly doubles its share in work tasks compared to personal ones. ChatGPT and Gemini are generalists tilted toward informational and everyday use; the smaller platforms have carved out defensible niches that their overall share figures do not reveal.

The implication for brands optimising for AI visibility is significant. A brand that tracks only ChatGPT and Gemini is measuring the majority of the market but missing the task-specific contexts where other platforms dominate. A developer tool brand, for instance, may find that Claude is the more consequential platform for its category — even though Claude's headline share is a fraction of ChatGPT's.

<h2>Trust is earned, not inherited</h2>

The study's most striking finding on trust concerns the gap between reputational and experiential trust. ChatGPT and Gemini are already trusted by people who have never used them — 54% and 60% of aware non-users rank them among their three most trusted platforms, riding the brand recognition of OpenAI and Google. Claude is different. Only 41% of those aware of it but not using it rank it top-three, but that figure rises to 76% among its actual users: a 35-point experiential lift, nearly 1.5 times ChatGPT's 24-point lift and the largest of any platform measured.

In head-to-head comparisons restricted to users who have experience with both platforms, Claude is ranked above ChatGPT (59–41) and above Gemini (66–34). ChatGPT is ranked above Gemini (60–40). The ordering is consistent and statistically clean. The competitive implication is that Claude's reputation understates how its actual users regard it, while the incumbents' reputations roughly match or exceed their users' experience. For brands, this matters because the platform a user trusts most is likely the one whose recommendations carry the most weight.

<h2>The privacy paradox is an information problem</h2>

More than 80% of users report concern about how their conversation data is used, yet roughly 60% do not know whether their assistant trains on their conversations, and only 18% have ever paid for a plan with better privacy protection. The dominant predictor of protective behaviour is not concern but policy literacy: knowing whether one's assistant trains on conversations is associated with 0.34 additional protective tools adopted — the largest effect in the regression model and larger than a full point of concern on the five-point scale. Concern matters, but its coefficient is roughly a third the size of literacy's.

The study reframes the privacy paradox as an information problem rather than a preference problem. Users are not indifferent to privacy; they are uninformed about the specific practices that would prompt action. Disclosure and defaults, the paper argues, are what would move protective behaviour — not campaigns designed to raise concern.

<h2>What users actually pay to protect</h2>

The discrete-choice experiment asked users to choose between hypothetical AI plans that varied on monthly price and three data-handling attributes: human review of conversations, use of conversations for model training, and sponsored answers. The results reveal a steep hierarchy. Users pay most to avoid human review of their conversations — $11.20 per month — nearly four times what they pay to avoid training data use ($2.97), with avoiding sponsored answers in between ($6.46). Valuations rise with task sensitivity: for highly sensitive tasks, the willingness to pay to avoid human review reaches $15.24 per month.

The finding inverts the public debate. Training data use dominates regulatory and media attention, but it is the feature users value least of the three. The prospect of a person — an employer, a contractor, a government requester — reading their words is what users are actually willing to pay to prevent. For AI platforms competing on privacy, the paper suggests that human-access guarantees may be a more commercially potent differentiator than training-policy commitments.

<h2>AEO Updates Takeaway</h2>

For brands and marketers thinking about AI visibility, this study offers three practical reframings. First, platform share at the aggregate level is a poor guide to platform importance at the task level — category-specific measurement matters. Second, the platform a user trusts most is not necessarily the one with the highest headline share, and trust is built through use rather than reputation, which means challenger platforms can earn disproportionate influence in specific categories. Third, the privacy conversation that AI platforms are having with regulators is not the same conversation their users are having with themselves — and brands that understand that gap will be better positioned to navigate the trust dynamics of AI-mediated recommendation.

<h3>References</h3>

[1] Zou, J. (July 2026). Platform Choice, Privacy, and Task Allocation in the Consumer AI Assistant Market. Profound.

Mainstream business and technology outlets have begun treating AI search as a structural economic shift, not a product feature. That transition in coverage signals a new stage of maturity for the category.

Mainstream business and technology outlets have begun treating AI search as a structural economic shift, not a product feature. That transition in coverage signals a new stage of maturity for the category.

For the past year, most discussions about Answer Engine Optimisation have taken place inside marketing departments, SEO communities, and AI startups. That may be changing. Over the past two weeks, mainstream media outlets have started treating AI search as an economic transformation rather than a technology story. That is an important milestone, not because AEO suddenly became real, but because the broader business community is beginning to realise that it is.

<h2>The story has changed</h2>

Until recently, most headlines about AI search focused on product launches: ChatGPT added search, Google launched AI Overviews, Perplexity gained users, Anthropic improved Claude. Those stories mattered, but they largely treated AI search as another feature in the AI race. The newest coverage asks a much bigger question: what happens to the web when people stop clicking links and start accepting answers? That is no longer a product story. It is an economic one.

<h2>From a battle for clicks</h2>

For more than twenty years, digital marketing revolved around one objective: get the click. SEO, paid search, content marketing, and analytics were all designed around moving users from Google to a website. AI search changes that equation. Increasingly, the answer itself is the destination, and the website becomes the supporting evidence. As Le Monde put it, the competitive landscape is shifting from a battle for clicks to a battle for inclusion in AI-generated answers. That single observation captures why AEO exists.

<h2>Publishers are feeling the pressure</h2>

The conversation is no longer theoretical. Publishers are openly discussing declining referral traffic, changing audience behaviour, and the need to rethink their business models as AI-generated answers reduce the number of traditional website visits. For years, marketers asked how to rank. They are increasingly asking how to become one of the sources AI chooses to trust. Those are very different questions, and the shift in framing reflects a genuine change in what the competitive landscape demands.

<h2>The next stage is already emerging</h2>

Perhaps the most consequential idea in the recent coverage is the concept of an agentic web. Rather than simply answering questions, AI systems may increasingly book appointments, compare products, make purchases, and complete workflows on behalf of users. If that happens, organisations will not simply be optimising for human visitors; they will be optimising for AI agents acting on behalf of those visitors. That has implications far beyond marketing, touching product discovery, e-commerce, customer service, digital interfaces, and ultimately how businesses compete online.

<h2>What this means for brands</h2>

The question is no longer whether AI search matters. That debate is largely over. The better question is how organisations should respond. Many companies are understandably focused on visibility, which is an important first step, but visibility alone will not be sufficient. Brands will increasingly need to understand why AI recommends certain companies, which claims AI consistently repeats, which sources shape those claims, how those perceptions evolve over time, and what interventions actually change AI's understanding of a brand. Those questions extend beyond traditional SEO and even beyond basic AEO.

<h2>AEO Updates Takeaway</h2>

The biggest story is not that AI search is growing. The biggest story is that mainstream business and technology media have started treating AI search as a structural shift in the economics of the web. When the conversation moves from announcing a new AI feature to describing how the internet itself is changing, a category has reached a different stage of maturity. That does not mean all the winners have been decided. It means the rest of the market is finally starting to pay attention, and when that happens, the conversation usually shifts from whether a new category matters to who will define it.

One of the world's leading AI measurement companies is investing heavily in out-of-home advertising. The apparent contradiction is actually the most coherent thing in marketing right now.

One of the world's leading AI measurement companies is investing heavily in out-of-home advertising. The apparent contradiction is actually the most coherent thing in marketing right now.

Trevor Pyle, Head of Marketing at Profound, published a LinkedIn post this week that had nothing to do with AI search. It was about billboards. Specifically, why Profound is investing heavily in out-of-home advertising despite openly acknowledging that it is one of the least measurable marketing channels available. Profound, of course, helps brands understand how they are represented in AI search. The apparent irony is worth examining, because it turns out to be no irony at all.

<h2>Measuring signals, not certainty</h2>

Trevor's list of success metrics for the billboard campaign includes salespeople feeling more confident, job candidates mentioning the billboards, investors recognising the brand, friends texting photos, and competitors asking whether they should buy billboards too. These are not vanity metrics. They are signals of a kind of influence that does not resolve cleanly into a dashboard. The campaign is not being evaluated by attribution; it is being evaluated by the quality of the conversations it generates.

<h2>Marketing has always included things we could not fully measure</h2>

Long before AI search existed, marketers were already making investments they could not perfectly attribute. Brand, PR, events, podcasts, word of mouth, executive thought leadership, and out-of-home advertising all create effects that show up somewhere else. The billboard does not close the deal. It changes how people think before the deal ever begins. That dynamic is not a failure of measurement; it is a feature of how influence actually works at scale.

<h2>AI search may be similar</h2>

One of the more persistent mistakes in AEO thinking is assuming that AI visibility exists in isolation from everything else a brand does. It does not. AI models are trained on the internet, and the internet is shaped by people who create articles, research, discussions, citations, recommendations, links, and reviews. Brand building still matters, and it may matter more than ever, because the signals that shape AI representation are the same signals that have always shaped reputation. The channels have changed; the underlying logic has not.

<h2>The measurement challenge is not new</h2>

Trevor's post is really about something larger than billboards. It is about accepting that some of marketing's highest-return activities leave imperfect measurement trails. That does not make them unimportant. It simply changes how success is evaluated. The same is becoming true in AI search. Not every improvement in AI representation will be traceable to a single article, a single prompt, or a single optimisation. Sometimes dozens of small signals accumulate until the models begin representing a brand differently. The challenge is not eliminating uncertainty. It is building better ways to observe it.

<h2>AEO Updates Takeaway</h2>

Marketing has spent two decades chasing perfect attribution. AI search may be a reminder that influence has always been more complicated than a dashboard. Measurement will continue to improve, frameworks will become more sophisticated, and experiments will become more rigorous. But some of the most valuable things brands create, including trust, reputation, authority, and familiarity, will probably remain only partially measurable. There is something clarifying about one of the world's leading AI measurement companies buying billboards. Not because it is contradictory, but because it is consistent. The people building the future of measurement also understand the enduring value of things that cannot be perfectly measured.