- AI visibility score already lists the questions worth asking before trusting a score. This piece applies them directly, vendor by vendor, to three real, funded companies in the category, and the gap between what they each disclose is large.
- Ahrefs' Brand Radar is the most disclosed of the three checked here: a stated 421M+ prompt corpus, per-engine monthly volumes broken out individually, a monthly update cadence, and a searchable raw corpus rather than just a summary number.
- Semrush's AI Visibility Index discloses a headline number (126M prompts, Jan-Apr 2026) and a general description of its inputs, but has drawn specific criticism for not disclosing its underlying data sources and methodology in comparable depth to Ahrefs.
- Profound states a larger raw dataset than either (1.5 billion-plus real user AI conversations, 100M+ queries processed monthly), but per independent reviews of the platform, doesn't publish comparable technical documentation of how that data becomes a score, which is a separate axis from dataset size entirely.
- Dataset size and methodology transparency are not the same thing, and a vendor's funding or market position is not a proxy for either. Check the actual disclosure, not the reputation.
AI visibility score lays out four questions worth asking before trusting any score: how many prompts, how many runs per prompt, over what window, and blended or per-platform. That piece keeps the answer general on purpose. This one applies those questions directly to three real, currently-operating, funded vendors in the category, not to argue one is better than another as a product, but to make the actual gap in what each one discloses concrete and checkable.
How disclosed is Ahrefs Brand Radar's methodology?
Ahrefs states its Brand Radar corpus at more than 421 million organic prompts, with monthly volumes broken out per engine individually: ChatGPT and Copilot at roughly 13.3 million each, Perplexity at 13.3 million, Gemini at 12.4 million, AI Overviews at 143 million, and AI Mode at 41 million. The corpus updates monthly, and (distinct from the other two vendors here) Ahrefs makes the raw response corpus itself searchable directly, rather than only surfacing a summary score derived from it. A customer can go look at what was actually asked and what was actually returned, not just trust the number that came out the other end.
This doesn't mean Ahrefs' methodology is beyond question: a monthly-refresh cadence is slower than real-time, and per-engine query-volume figures are still Ahrefs' own disclosed numbers, not independently audited by a third party. But relative to the other two vendors checked here, it's the clearest case of "here's what went in, here's how much, here's where you can check it yourself."
How disclosed is Semrush AI Visibility Index's methodology?
Semrush states its 2026 AI Visibility Index scales to an analysis of roughly 126 million US AI search prompts, collected January through April 2026, built by combining keywords from its own search module with AI-generated questions fed into major LLMs. That's a real, specific number and a general description of the method, meaningfully more than a vendor disclosing nothing. But it stops short of Ahrefs' level of granularity: no comparable per-engine volume breakdown has been located, and the platform has drawn specific criticism for not disclosing its underlying data sources and methodology in the same depth. A customer gets the headline number and a general sense of the inputs, without the same ability to inspect the underlying corpus that Ahrefs provides.
How disclosed is Profound's methodology?
Profound states its dataset at more than 1.5 billion real user AI conversations, with more than 100 million queries processed monthly, a larger raw scale, by its own stated figures, than either of the other two vendors here. Profound also states that it runs tracked prompts daily and captures responses directly from the browser interface, aiming to reflect what an actual user would see rather than a synthetic API call.
Here's where the two axes genuinely separate: per independent reviews of the platform, Profound does not publish the same depth of technical documentation about how that raw data becomes a visibility score, a rank, or a share-of-voice figure, making the underlying methodology harder to independently validate than the scale of the claimed dataset alone would suggest. A bigger number of conversations analyzed says nothing about whether the scoring methodology built on top of it is disclosed, auditable, or comparable to a competitor's. This is also the best-funded vendor of the three checked here ($154.5 million raised across four rounds in roughly eighteen months), which is worth noting specifically because funding size and methodology transparency turned out to be unrelated in this comparison, not correlated.
How do the three compare, side by side?
| Vendor | Stated dataset | Per-engine breakdown | Raw data inspectable | Update cadence |
|---|---|---|---|---|
| Ahrefs Brand Radar | 421M+ prompts | Yes, disclosed per engine | Yes, searchable corpus | Monthly |
| Semrush AI Visibility Index | ~126M prompts (Jan-Apr 2026) | Not disclosed at comparable granularity | No | Not specified |
| Profound | 1.5B+ conversations, 100M+ queries/month | Not disclosed at comparable granularity | No | Daily (per prompt tracked) |
Why does this comparison matter more than picking a "winner"?
The point isn't that Ahrefs is the correct vendor to use: different tools serve different use cases, and a larger real-conversation dataset (Profound's stated advantage) or a different prompt-generation method (Semrush's blend of search-module and AI-generated queries) could genuinely be the better fit for a specific question. The point is narrower and more useful: dataset size, funding, and market position are three separate things from methodology transparency, and none of them substitutes for it. A buyer evaluating any vendor in this category (including ones not covered here) should ask for the specific disclosure Ahrefs provides (prompt count, per-engine breakdown, an inspectable corpus, a stated update cadence) rather than accepting a large headline number as proof the underlying methodology is sound.
Frequently asked questions
Which AI-visibility vendor discloses the most about its methodology?
Of the vendors checked for this piece, Ahrefs' Brand Radar discloses the most: a stated 421-million-plus prompt corpus with per-engine monthly volumes broken out individually (ChatGPT and Copilot at roughly 13.3 million each, Perplexity 13.3 million, Gemini 12.4 million, AI Overviews 143 million, AI Mode 41 million), a monthly update cadence, and a raw corpus a customer can search directly rather than just a summary score.
Does a bigger claimed dataset mean better methodology transparency?
No. The two are separate things, and conflating them is a common mistake. Profound has publicly stated a dataset of 1.5 billion-plus real user AI conversations and processes over 100 million queries a month, a larger raw scale than the other vendors checked here. But per independent reviews of the platform, it does not publish the same level of technical documentation about how that data becomes a score, which makes the underlying methodology harder to independently validate regardless of the dataset's size.
Does Semrush disclose its AI Visibility Index methodology?
Partially. Semrush states its 2026 AI Visibility Index draws on roughly 126 million US AI search prompts collected January through April 2026, combining its own search-module keywords with AI-generated questions fed into major LLMs. That's more disclosure than a vendor stating no numbers at all, but it falls short of Ahrefs' level of detail; Semrush has drawn specific criticism for not disclosing its underlying data sources and methodology in comparable depth.
What should I ask any AI-visibility vendor before trusting their number?
The same four questions from AI visibility score, applied vendor by vendor: how many prompts, are runs repeated enough per prompt to be reliable, over what time window, and is the score blended across platforms or broken out per engine. A vendor's size, funding, or market position says nothing about whether it answers these four questions. Check the actual methodology page, not the vendor's reputation.
Sources referenced
- Ahrefs: Brand Radar methodology (421M+ prompt corpus, disclosed per-engine volumes)
- Semrush: 2026 AI Visibility Index (126M US AI search prompts, Jan-Apr 2026)
- Profound: Prompt Volumes feature page
This piece is the vendor-by-vendor application of the checklist in AI visibility score. For the funding and incentive layer behind why this transparency gap persists, see the AI visibility score industry: who's really being served; for whether any of these scores predict future outcomes at all, see do AI visibility audits actually predict anything?
