AI Visibility
See how your brand and competitors appear in the answers your project measures.
Use AI Visibility to investigate brand presence across a defined set of prompts and engines. Start with the selected run and filters, then move from the summary into the answer evidence.
Run a useful audit
Choose the prompts, logical engines and repetitions in the audit launcher. Review the estimate before confirming. More repetitions provide more observations under that run’s conditions; they do not turn a sample into a guarantee about every user’s experience.
An audit keeps the prompts, tracked roster and measurement configuration it used. Check Runs for completion, cancellation and unsuccessful attempts.
Read Trends
Trends is the default view. Select the run or period, engine and cohort appropriate to your question. When using a baseline, check whether the comparison is compatible.
Keep these observations separate:
| Observation | What it tells you |
|---|---|
| Brand mention | The answer mentions a recognized brand |
| Owned citation | The answer cites a URL belonging to an owned domain |
| Recommendation | The answer contains a recognized recommendation |
| Position | A supported ordering places an entity at a particular rank |
A citation is not automatically a brand recommendation. Unsupported assessments remain unavailable.
Inspect Sources
Switch to Sources to see the domains and URLs used in the selected evidence. Open a domain to inspect its URLs and the prompts that reached it. Open an answer to check how a citation was used.
See Sources and citations for interpreting retrieval, citation rates and brand co-occurrence.
Read Query Fanout in context
Query Fanout shows captured query evidence associated with the measured answers. Use it to understand the recorded retrieval context. Absence of captured fanout does not prove that an engine did no research.
Respond to a change
First check whether the engine, prompt set, cohort or successful-answer coverage changed. Then inspect which answers and sources explain the observed difference.
Use an Action or a focused Agent question to turn the evidence into work. Avoid attributing movement to a website edit until you have examined later compatible measurements and their limitations.