How Prescy measures AI visibility

What Prescy checks

A full visibility scan sends your approved, tracked buyer searches to ChatGPT, Perplexity, Gemini and Claude through their supported integrations. It also checks Google AI Overviews through Google search results. Prescy stores the available answer for each search and engine so you can inspect the evidence behind the result.

Google AI Overviews are currently checked using US English search results. Models, provider interfaces and supported surfaces can change; Prescy may update an integration to keep the measurement working.

The four chat engines use the full search sample allowed by your plan. Because Google AI Overviews is substantially slower, a weekly full scan currently checks up to 3 searches there, and a priority scan checks up to 1. The exact Google sample is displayed with each result.

The searches being measured

During setup, Prescy suggests buyer-style searches based on your website. You review and can edit them before the first scan. These are intended to represent questions a potential customer might ask while discovering, comparing or choosing a product or service.

What counts as a brand mention

A result counts as a mention when the available answer contains your brand name, a saved brand alias, your full domain or an unambiguous domain label. Matching is case-insensitive and uses word boundaries so a short brand name is not counted inside an unrelated word.

Per-engine visibility

Each engine's visibility percentage is the number of applicable answers that mention your brand divided by the number of applicable answers successfully checked for that engine.

Example: if your brand appears in 7 of 20 available ChatGPT answers, ChatGPT visibility is 35%.

The 0–100 score

The overall score combines two signals:

• Breadth, weighted 41% — the share of scorable engines that mention your brand at least once.
• Depth, weighted 59% — the average per-engine visibility across all scorable engines, including zeroes.

For example, appearing on 3 of 5 engines with engine visibility of 60%, 40%, 20%, 0% and 0% gives 60% breadth and 24% average depth. The resulting score is 39.

Missing, delayed and unavailable results

• An engine that could not run is marked unavailable and excluded from the score.
• Google AI Overviews may finish after the chat engines. While it is pending, it is not included; the score updates when Google completes.
• If Google was checked but did not show an AI Overview for a search, that search is not treated as a brand absence. If no tracked search showed an AI Overview, Google is excluded from that score.
• If only some requests succeed, the engine percentage uses the successfully available answers as its denominator.

The dashboard displays which engines and how many searches were included, so two scans with different coverage can be interpreted correctly.

Competitor measurement

Competitor results combine appearances in your tracked buyer searches with bounded, paid-plan comparison searches. Those comparison searches include head-to-head and alternatives questions. A competitor being named in the question itself does not count as visibility; Prescy looks for its appearance in the answer.

Competitor queries are stored separately and never change your own brand's score. Competitors are matched by the name you provide, so aliases that have not been entered may be missed.

How recommendations are produced

After measurement, a reasoning model reviews the collected answers to identify likely winners, gaps and prompt-specific actions. It is instructed to use only the supplied answers. Evidence quotations are checked against the actual source text before being marked verified, and the verbatim engine answers remain available for your own review.

Recommendations are generated analysis, not guaranteed outcomes. They should be reviewed alongside your customer knowledge, positioning and wider marketing data before implementation.

How recommendation outcomes are measured

When you mark an action implemented, Prescy keeps the full scan that produced it as an immutable baseline. Later results only count when a full scan contains the exact same tracked customer search. The first qualifying rescan is labelled an early signal; two consecutive results pointing in the same direction are labelled stronger evidence.

Improved timing does not prove the action caused the movement. AI answers can also change because of model updates, new sources, recrawling and normal variation. Prescy therefore reports improved, no clear change, declined or mixed—not guaranteed attribution.

Free recognition check versus full visibility scan

The free website checker answers a narrower question: does AI recognise this brand? It asks three fixed, brand-focused questions across ChatGPT, Claude, Gemini and Perplexity and reports how many available engines recognised the domain.

The full account scan answers a different question: does AI surface the brand when buyers search for its category, use case or alternatives? It uses your approved buyer searches and produces the 0–100 visibility score. Recognition and buyer visibility should not be compared as though they are the same measurement.

Limits and interpretation

• AI answers are probabilistic and can vary between otherwise identical requests.
• The score measures the selected searches, engines and available answers—not every possible customer question or AI product.
• A higher score does not prove preference, sentiment, accuracy, traffic or revenue.
• Provider model changes can create movement even when your website has not changed.
• Small moves of approximately 1–2 points are treated as normal measurement variance; trends over several scans are more meaningful.