How CompetLab measures AI visibility
How we collect
We collect AI-visibility data by putting real, buyer-style questions to the AI engines and reading what comes back — the same way a prospective customer would ask an assistant to recommend a tool in your category. Nothing is scraped from a search page; the measurement is the answer an engine actually gives.
Each project has a set of monitoring prompts — the questions a buyer might ask about your space. For a check, CompetLab sends each prompt to three engines: ChatGPT (OpenAI), Claude (Anthropic), and Gemini (Google). In the standard setup that is three prompts across three engines, or nine queries per check. We then read every answer for brand mentions — your own domain and each tracked competitor — recording whether a brand is named and in what order it appears.
Your monitoring prompts
│ (buyer-style questions about your space)
▼
Ask each engine → ChatGPT · Claude · Gemini
│ 3 prompts × 3 engines = 9 queries per check
▼
Read every answer for brand mentions
→ who is named (you + each competitor)
→ in what order (rank; lower is better)
▼
Roll up per brand
→ Mention Rate → average rank → per-engine split
→ AI Visibility Score (0–100 composite)As the diagram shows, the same nine-query cycle runs for your brand and every competitor, so the numbers are comparable across the whole set. From the parsed answers we compute a Mention Rate (the share of queries that name a brand), an average rank (its ordinal position when it does appear), a per-engine breakdown, and the composite AI Visibility Score. An agent can pull exactly these fields through the MCP tools reference or the AI Visibility endpoints.
How we compute the AI Visibility Score
The AI Visibility Score is a single weighted composite from 0 to 100, where higher is better. It summarizes two things our data captures separately: how often a brand is mentioned across the nine queries in a check (Mention Rate), and how prominently it ranks when it does appear (average rank, where lower is better).
The score is computed per brand — your domain and each competitor each get one — and it spans all three engines at once. It is deliberately not a per-engine number. That single cross-engine design is why a provider-filtered trend returns no score at all: there is no separate ChatGPT score or Claude score to hand back, only the combined one.
We do not publish an exact public weighting formula, and we won’t invent one here. What we can be precise about is that the raw inputs behind the score are all returned by the API: per-brand mention counts, the Mention Rate, the average rank, and the per-engine (openai, claude, gemini) breakdown are exposed alongside the composite. So even though the internal weighting is ours, you can audit the underlying evidence for any score in the AI Visibility endpoints rather than taking the number on faith.
Refresh cadence
Refresh cadence depends on the dimension: the five monitored dimensions run continuously on a schedule, while the once-a-month Strategic Briefing carries the deeper read. AI Visibility is a monitored dimension, so your score is refreshed on a recurring cycle rather than only when you ask for it.
| What | How often | History |
|---|---|---|
| Monitored dimensions (AI Visibility, Positioning, Pricing, Content, Tech & Trust) | A recurring schedule, measured in days | Every run and check is retained; page the history or chart the trend |
| Researched dimensions (the other eight) | Refreshed with each briefing edition | Surfaced in the briefing — no standalone run history |
| Strategic Briefing | Roughly every 30 days | Only the latest finished edition is returned |
Because every monitored run is kept, day-to-day movement is visible as a trend, not just a latest value — you can page through the check history or read a time series. The eight researched dimensions are refreshed as part of the Strategic Briefing, which regenerates roughly every 30 days; see the strategic-briefing endpoint for the availability states and dating. For how each monitored dimension is scheduled, see Monitoring.
What we do not claim
AI answers are probabilistic, so we are deliberate about what these numbers do and do not prove. Being explicit about the limits is part of the methodology, not a disclaimer bolted on to it.
- Answers vary run to run. The same prompt can produce a different answer on a different day. A single check is a snapshot; the trend over time is the reliable read, which is exactly why we sample on a schedule and keep history.
- We measure a sample, not the whole universe. A check runs your configured prompts — a representative set of buyer questions — not every way a person could phrase one.
- We measure three AI engines. ChatGPT, Claude, and Gemini. We don’t claim to cover every model, assistant, or answer surface, and we don’t model how one logged-in user’s personalization or region might change a reply.
- Correlation, not causation. A score change lines up with what you and your competitors do, but it doesn’t prove why it moved. Treat it as a signal to investigate, not a verdict.
- We measure; we don’t control. CompetLab observes what the engines say. It doesn’t inject content, pay for placement, or steer a model.
- Conveniences aren’t ranking levers. Things like an
llms.txtfile are forward-looking niceties for coding and RAG agents. We don’t claim they are proven AI-ranking or citation factors, because they aren’t.
Authorship
This methodology is maintained by the CompetLab team that builds and operates the measurement pipeline described above. When the pipeline changes — the engines we query, how a check is composed, how the score is built — we update this page and the affected references together, so the docs don’t drift from what the product actually does.
If a figure here ever looks inconsistent with what the API returns, the API response is the authoritative source; tell us and we’ll reconcile the page. This page was last reviewed on 10 July 2026.
Related
Where these numbers live, and how to pull them yourself.
FAQ
How does CompetLab measure AI visibility?
CompetLab measures AI visibility by asking the major AI answer engines real, buyer-style questions and reading their answers. For each project we take a set of monitoring prompts — the kinds of questions a buyer would ask about your space — and send each one to ChatGPT, Claude, and Gemini. In the standard setup that is three prompts across three engines, or nine queries per check. We then parse every answer for brand mentions: whether your brand and each competitor is named, and in what order. Those results roll up into a Mention Rate, an average rank, per-engine breakdowns, and a single AI Visibility Score, computed the same way for you and every competitor so the picture is comparable.
Which AI engines do you query?
CompetLab queries three AI answer engines for AI Visibility: ChatGPT (OpenAI), Claude (Anthropic), and Gemini (Google). Every check runs your prompts across all three, and the results are reported both per engine and as one cross-engine score, so you can see where a gap comes from. We name only the engines we actually query — we don't claim to cover every model, assistant, or answer surface, and coverage can change as the pipeline evolves. Because these engines personalize and localize their answers, what we measure is a consistent, repeatable sample rather than a guarantee of what any single logged-in user will see on a given day.
How is the AI Visibility Score calculated?
The AI Visibility Score is a single weighted composite from 0 to 100, where higher is better. It summarizes two things our data captures separately: how often a brand is mentioned across the nine queries in a check (Mention Rate) and how prominently it ranks when it does appear (average rank, where lower is better). The score is computed per brand and spans all three engines at once — it is not a per-engine number, which is why a provider-filtered trend returns no score. We don't publish an exact public weighting formula, but the raw inputs behind the score — mention counts, ranks, and per-engine breakdowns — are all returned by the API, so you can audit exactly what went into it.
How often do you refresh the data?
It depends on the dimension. The five monitored dimensions — AI Visibility, Positioning, Pricing, Content, and Tech & Trust — run on a recurring schedule measured in days, and every run is kept in a history you can page through and chart as a trend. AI Visibility is one of those monitored dimensions, so your score is refreshed on a recurring cycle rather than only on demand. The eight researched dimensions are refreshed as part of the Strategic Briefing, which regenerates roughly every 30 days. So day-to-day movement shows up in the monitored trends, while the deeper, once-a-month read arrives with each new briefing edition.
How accurate is it, and how do you handle variance?
We treat any single check as a snapshot, not ground truth, because AI answers vary from run to run. The same prompt can produce a different answer on a different day, so a one-off number is less reliable than the trend behind it. That is why we sample repeatedly on a schedule, keep full history, and report both per-engine and cross-engine results — the direction and consistency over time is the signal to trust, not a single reading. The per-engine breakdown also shows where variance is coming from, so a swing driven by one engine is easy to spot. We would rather be honest about this than imply a precision that probabilistic systems can't offer.
Do you influence or only measure AI answers?
CompetLab only measures — it does not influence what the engines say. The pipeline observes AI answers; it does not inject content into a model, pay for placement, or otherwise steer a response. Nothing about measuring your visibility writes to or changes a third-party model. Improving your standing is your work, not ours: the score and the surrounding dimensions show you where you are absent or ranked low, and the Strategic Briefing suggests where to focus, but the changes happen in your own content, positioning, and product. We think that separation matters — a measurement you can trust has to be independent of the thing it measures.
What are the limits of this methodology?
The main limits are variance, sampling, engine coverage, and causation. AI answers are probabilistic, so results move run to run; we measure a configured sample of prompts, not every possible question a buyer could ask; and we query three AI engines, not every model, surface, or region. A score change correlates with what you and your competitors do, but it doesn't prove why it moved — treat it as a signal to investigate, not a verdict. We also don't claim that conveniences like an llms.txt file are proven AI-ranking or citation levers; they are forward-looking niceties, not established ranking factors. Being clear about these limits is part of the methodology, not a footnote to it.