AI visibility monitoring
One measurement is a snapshot. Monitoring is the record.
AI answers change. Monitoring reruns the same locked question set under comparable conditions so movement can be observed rather than assumed.
What monitoring tracks
- Recommendation share
- How often the team is recommended across the locked question set.
- Citation share
- How often the team or its pages are cited in the captured answers.
- Competitor share
- Which teams are recommended instead, and how that distribution moves.
- Question-level performance
- Which questions the team appears in, tracked question by question.
- Source changes
- New, lost or changed sources appearing around answers in the market.
- Entity accuracy
- Whether name, brokerage, markets and members are described correctly.
- Provider and date
- Recorded with every captured answer, because conditions change.
- Answer volatility
- How much the answers to the same question vary between runs.
How the two headline metrics are defined
| Metric | Definition |
|---|---|
| Recommendation share | Prompts where the team is recommended ÷ total valid prompts in the captured set. |
| Citation share | Prompts where the team or its pages are cited ÷ total valid prompts in the captured set. |
These are measurement metrics for a defined market, question set and provider. They are not universal rankings and they do not describe how any model ranks sources.
Invalid responses — refusals, non-answers, answers about a different market — are excluded from the denominator and reported separately, so a shrinking denominator can never quietly inflate a share.
What this cannot prove
- Movement between runs may reflect model changes rather than anything you did.
- Comparable conditions are maintained deliberately, but they cannot be guaranteed by us.
- No monitored metric predicts future answers.
Keep reading
See who AI recommends in your market — and where you stand.
One client per market. Start by telling me where you work.