Metrics and definitions
What we measure, in plain terms.
Every number in an AI Visibility Report is defined here: what it tells you, how to read it and what moves it.
Core metrics
How visible you are.
AI Visibility Score
- What it tells you
- A single number from 0 to 100 summarizing how visible your brand is in AI answers. It combines four signals: whether you are mentioned, whether you are cited, whether you are recommended, and how you are described.
- How to read it
- Higher is better. Use it to track progress over time and to compare pillars and AI assistants. The four signals do not count equally.
- What moves it
- Improving any of the four signals below it.
Mention Rate
- What it tells you
- How often AI names your brand when buyers ask unbranded questions in your market.
- How to read it
- Shown as a percentage of answers. A low rate on a pillar means AI does not associate you with that topic yet.
- What moves it
- Clear, specific content on the topic, and being discussed by sources AI relies on.
Citation Rate
- What it tells you
- How often AI links to or references your own website as a source for its answer.
- How to read it
- Shown as a percentage of answers. You can be mentioned without being cited, and cited without being recommended.
- What moves it
- Pages that answer the question directly, are easy for machines to read, and are trusted enough to quote.
Recommendation Rate
- What it tells you
- How often AI actively suggests your brand as a choice, as opposed to listing it in passing.
- How to read it
- Shown as a percentage of answers. This is the signal closest to a buying decision.
- What moves it
- Evidence AI can point to: reviews, comparisons, awards, clear product strengths.
Sentiment
- What it tells you
- Whether AI describes your brand in positive, neutral or negative terms, and the themes it attaches to you.
- How to read it
- Shown as a rating with example phrases. Neutral is common; the detail is in the themes.
- What moves it
- What reviews, press and your own site say about you, since that is what AI summarizes.
Position
- What it tells you
- Where your brand appears within an answer when several brands are named.
- How to read it
- Earlier is better. Being named first carries more weight with readers than being named fifth.
- What moves it
- Strength of association with the topic compared with the other brands named.
Context metrics
Where, and compared with what.
Pillar score
- What it tells you
- Any of the metrics above, calculated for one pillar on its own.
- How to read it
- Compare pillars to see where you are strong and where you are absent. An overall score can hide both.
- What moves it
- Work focused on that pillar.
Search versus AI gap
- What it tells you
- The difference between how strong you are in search for a pillar and how visible you are in AI answers for it.
- How to read it
- A large gap with search ahead is an opportunity. A gap with AI ahead is a strength to protect.
- What moves it
- Closing the content and markup issues that stop search authority carrying over.
Source mix
- What it tells you
- Where AI gets its information in your market: your own site, competitor sites, or third parties.
- How to read it
- A small share for your own site means others are telling your story.
- What moves it
- Earning a place on the third-party sources AI cites, and strengthening your own pages.
Share of voice
- What it tells you
- Your portion of all brand mentions in your market, compared with named competitors. Included with Competitor Analysis.
- How to read it
- Shown as a percentage. It answers "of the times AI names anyone, how often is it us?"
- What moves it
- Your own visibility, and your competitors'.
Structured data health
- What it tells you
- How complete and correct the JSON-LD is on your key page types.
- How to read it
- Shown as present, incomplete or missing for each page type and schema.
- What moves it
- Adding and correcting markup, which the report supplies.
Terms we use
A short glossary.
- A
AI assistant
A product people ask questions of and get written answers from, such as ChatGPT, Gemini, Perplexity, Claude and Google's AI Overviews.
- B
Unbranded question
A question a buyer would ask without naming you, such as "what are the best waterproof hiking boots?" This is the real test of visibility.
- C
Pillar
A topic your brand is associated with. Owned pillars are your brand and products. Category pillars are the wider themes you compete in.
- D
Mention, citation, recommendation
A mention is your name appearing. A citation is your website being used as a source. A recommendation is AI suggesting you as a choice.
How we keep it honest
Consistent, so you can compare.
We ask a broad set of real buyer questions, the same way, across every AI assistant, and score every answer by the same rules. A re-run uses the same questions and rules, so a change in your score reflects a change in what AI says, not a change in how we measured.
AI answers vary from one moment to the next, so we measure across many questions and report patterns, not single answers.
We publish what each metric means. The question sets, scoring rules and weightings behind them are our own method and are not published.