AI visibility

AI visibility, measured where it decides a purchase

AI visibility is whether an AI assistant names, cites and recommends your brand when someone asks a question that leads to buying. It is measurable, it differs sharply between engines, and it is not the same thing as being mentioned a lot.

Start Founding Agency Pilot — $499 Brand Pilot — $149

Founding pilot pricing while we validate. 14-day pilot, founder-run, cancel any time.

One question, three engines

Illustrative example

“Which [category] platform should a mid-size retailer use?”

ChatGPT your brand is put forward as the choice Recommended

Gemini named, not recommended Mentioned only

Perplexity Competitor A recommended Competitor wins

A single averaged “AI visibility score” would report this as 50% and hide the only fact worth acting on.

The three measurements that matter

Mentioned

Your brand name appears in the answer text. This is the weakest signal — an answer can name you while steering the buyer elsewhere, and a name in a list of nine is not a recommendation.

Cited

The assistant used a page you own as a source. This one is different in kind: it means your content is inside the reasoning, not just in the output. It is also the state you have the most direct influence over.

Why per-engine reporting is not a detail

The three engines ground their answers differently: OpenAI's web search tool, Google Search grounding for Gemini, and Perplexity's own retrieval. In our own measurements the same question routinely produces a different set of named brands on each. Averaging them into one "AI visibility score" destroys the finding that a buyer on one assistant sees you and a buyer on another does not.

So we never average across engines. Every result is reported per engine, with explicit coverage: an engine that failed to return a usable answer is marked unmeasured, never counted as evidence that you were absent.

How a run is recorded

Live capability

Per engine

Named · cited · recommended · which competitors appeared · which sources were used.

Per call

Whether the provider call actually succeeded. A failure is recorded as unmeasured.

Denominators

Failed calls are excluded. Absence of evidence is never scored as evidence of absence.

Raw answers

Kept, so a better detection rule re-scores history instead of re-purchasing it.

This is how the measurement engine we run today behaves — not a roadmap item.

What we found about stability

Before selling this we ran the same questions repeatedly to see what actually holds still. Whether a brand was recommended or cited was stable run to run. The exact set of competitors an engine names was not — it moved between runs.

That is why our reports show competitive share as directional: an ordering you can act on, not a decimal you should trust to two places. A tool that gives you a precise competitor share from a single run is reporting noise as signal.

Stable enough to report≠Precise

Recommendation status held; the competitor set drifted. Publishing the second as a two-decimal share would be dressing up run-to-run variance as a trend.

What we are still proving. That buyer-intent prioritisation beats generic visibility tracking is our working hypothesis, not a settled finding. We think measuring the questions that precede a purchase is more useful than counting mentions across a large prompt list — but we are testing that with paying customers, not asserting it. We have no customer results yet, and we will not invent any.

What we actually measure

Every result comes from the official OpenAI, Google Gemini and Perplexity APIs with web search or grounding enabled. In our data those surfaces are recorded as openai_api_web_search, gemini_api_google_search and perplexity_api_sonar, and reports name them explicitly.

That is not the same thing as a logged-in person using the ChatGPT app. These are official interfaces with web search on, but they are measurement surfaces rather than consumer sessions: no personalisation, no memory, no app-only features. Results can differ from what any individual user sees, and we do not claim the two are identical. We would rather put that on the pricing page than in the small print.

This page is the category overview. For the measurement model in detail, read AI search visibility; for the metric definitions underneath it, AI search analytics; for what the product does with them, the AI visibility tool.

What is AI visibility? · AI search monitoring · AI brand monitoring · AI visibility report · ChatGPT visibility · Answer engine optimization