Prompt set
We define buyer-intent questions across discovery, comparison, evaluation, use case, and category language. The set is agreed before baseline measurement.
Methodology
A repeatable sample, not a claim of universal truth.
AI answers are probabilistic, personalized, and frequently updated. Our method makes the inputs, observations, and limitations explicit so teams can make better decisions.
Platform sampling
We run prompts on the contracted platforms within a documented collection window and record answer evidence, sources, and relevant settings.
Human classification
Mentions, recommendations, citations, and competitor appearances are checked against consistent definitions, with ambiguous cases annotated.
Definitions
The metrics we report
| Metric | Definition |
|---|---|
| Mention rate | Share of sampled answers in which the brand is named, regardless of recommendation strength. |
| Recommendation rate | Share of relevant sampled answers in which the brand is positively suggested or shortlisted. |
| Citation rate | Share of sourced sampled answers that cite the brand’s owned domain. |
| AI share of voice | The brand’s share of qualifying appearances among the selected competitor set within the sample. |
Limitations
What the numbers do not mean
- A sample is not every possible prompt or every user.
- Outputs can differ by geography, language, account state, browsing mode, model, and time.
- Some interfaces do not expose complete source data.
- A change between samples may reflect model variation as well as brand-side work.
- Visibility does not automatically equal qualified traffic or revenue.
What the Free Visibility Snapshot gives you
We review your product, market, competitors, and most important buyer question. When a useful sample is possible, we return a concise observation and a sensible next step.