AI Visibility
How to measure AI visibility
A repeatable method: define the query set, sample systematically, and report measures separately.
AI visibility only means something relative to a stated method. Two reports with different query sets or systems cannot be compared.
1. Define the query set
Group queries by intent: definitional, comparative, decision and navigational. Include the questions customers actually ask. Record the full list; it is part of the result.
2. Choose systems and conditions
Name each AI system sampled, the date, the locale, and whether the session was signed out or personalised.
3. Sample more than once
Answers vary between runs. Repeat each query several times and report the share of runs, not a single outcome.
4. Report measures separately
| Measure | Question it answers |
|---|---|
| Mention rate | In what share of answers is the organisation named? |
| Citation rate | In what share is one of its pages cited? |
| Recommendation rate | In what share is it recommended for the task asked? |
| Entity accuracy | When described, are the facts correct? |
| Source diversity | Which sources are cited when it is mentioned? |
| Competitor share | How does it compare on the same queries? |
Combining these into one score hides the information you need to act on.
5. State the limits
Every report should say what it did not cover: systems not sampled, languages excluded, and how often results changed between runs.
Related reading
- Standards UpdatesGEO Standard v1.0: scope and design decisionsWhy the first version assesses ten areas, separates measures, and excludes guarantees.6 min read
- GEO ResearchThe limits of measuring AI searchVariance, personalisation and opaque systems: what any AI visibility number can and cannot tell you.5 min read
- AI SearchWhat is Generative Engine Optimization?A plain definition of GEO, what it changes about search work, and what it cannot promise.6 min read
Terms used here are defined in the glossary.