Ahrefs’ prompt index has passed 460 million. The methodology matters more than the headline

AEO platforms are beginning to compete on something more consequential than dashboard design. They are competing on the data layer that determines what a visibility score can see in the first place.

Ahrefs now says Brand Radar covers more than 463 million monthly prompts across AI Overviews, AI Mode, ChatGPT, Copilot, Gemini, Perplexity and Grok. That is a substantial stated scale. It is also a useful moment to ask the question behind every large AEO index: what, exactly, is being counted? [1]

<h2>A prompt index is the sample beneath the score</h2>

No platform can test every question a person might ask an assistant. The practical answer is to construct a finite prompt universe, run it through named systems, and code the resulting responses. That universe is the prompt index. It is the measurement sample beneath any resulting figure for visibility, share of voice, citations or recommendation presence.

The distinction matters because two platforms can produce different scores for the same brand without either figure being mechanically wrong. One may centre broad category questions. Another may include more comparison, problem, product-attribute, geographic or follow-up questions. Each is measuring a different denominator.

<h2>Ahrefs combines behavioural signals with semantic coverage</h2>

Ahrefs describes Brand Radar as anchored in questions drawn from its keyword data and Google’s People Also Ask corpus, then expanded with semantic fanout. In its methodology, Ahrefs frames the combination as a trade-off between behavioural relevance and topical completeness. The benefit is broad coverage. The limitation is equally important: search-derived questions are not a census of what people privately ask ChatGPT, Gemini or another assistant. [2]

That is not a flaw unique to Ahrefs. The true distribution of prompts inside most answer engines is not public. Every external measurement provider therefore has to make a modelling choice about which questions stand in for that hidden population.

<h2>Scale improves coverage, not representativeness by itself</h2>

A small prompt set can be overly sensitive to the questions chosen. A very large set can cover far more categories, constraints, comparison patterns and long-tail needs. But a large dataset can still overstate a result if it disproportionately captures questions that are commercially peripheral or weakly related to how a target audience actually makes decisions.

The useful buyer question is not simply whether an index is large. It is how the index is sourced, refreshed, weighted, localised and separated by intent. Ahrefs itself notes that its modeled visibility measures indicate potential exposure rather than actual audience reach. [3]

<h2>Discovery and control solve different problems</h2>

The broad index is useful for discovery. It can show where a brand appears across a large managed universe and surface patterns the team did not know to look for. Custom prompts serve a different purpose. They let a company track the defined questions, locations and refresh cadence that matter to a specific commercial strategy.

Those two layers should not be confused. The first helps reveal the landscape. The second makes it possible to manage a controlled measurement programme.

<h2>AEO Updates Takeaway</h2>

The data race is real, but the largest prompt index will not automatically be the most useful one for every buyer. A credible AEO measurement programme should expose its denominator, distinguish modeled visibility from audience measurement, and pair broad discovery with a transparent custom prompt register. The number on the dashboard is only as interpretable as the sample beneath it.

<h3>References</h3>

[1] Ahrefs. <a href="https://ahrefs.com/brand-radar" target="_blank" rel="noreferrer">Brand Radar</a>. Accessed 16 August 2026.

[2] Ahrefs. <a href="https://ahrefs.com/blog/brand-radar-methodology/" target="_blank" rel="noreferrer">Ahrefs Brand Radar Methodology: How we collect and model AI visibility data</a>. 26 February 2026.

[3] Ahrefs Help Centre. <a href="https://help.ahrefs.com/en/articles/11064852-what-is-brand-radar-and-how-to-use-it" target="_blank" rel="noreferrer">What is Brand Radar, and how to use it?</a>. 13 July 2026.