Method
What to Confirm in the Profile After a Site Scan
A site scan is only the starting point. Before you read any AI visibility number, confirm three things in the profile: what was observed, how missing values are marked, and that API responses stay separate from what a person sees on screen.
After a site scan, three things in the profile need confirming: the scope that was observed, how missing values are marked, and whether the sources stay separate. Open the profile and check that the scope matches the site you scanned. Check that any missing measurement is marked N/A and not zero. Then check that answers returned through an API are kept apart from what a person sees on a screen. Only once those checks are done is the profile ready to be read as a baseline for measuring AI visibility.
Why confirming the profile is the first step in measuring AI visibility
Any question about how AI visibility gets measured runs through the profile first. Before you compare periods or surfaces, you need to know what the profile is actually observing and how it handles what it does not observe. When a value is missing, RankAI shows it as N/A, never as zero. That is a product fact, and it belongs in the profile check: N/A is not a negative result, it is the absence of a measurement.
Sources are the second point. What comes back from an engine through an API is the API response, not the screen a person sees. The profile should confirm that the two stay separate. The same applies to Gemini and Google AI Overviews: they are not the same observation and should not be treated as a single source.
A working checklist from RankAI
The checklist below is a working proposal from RankAI, not a universal standard. Its job is to make the checks on a new profile explicit.
- Scope: the profile names the site that was scanned and the observation it refers to.
- Missing values: absent measurements appear as N/A, not as zero.
- Source separation: API responses are distinct from screens seen by a person; Gemini and Google AI Overviews are not merged into one entry.
- Reading a zero: if a zero appears, read it according to the semantics declared in the profile, not as an automatic absence of observation.
A RankAI reading proposal: treat zero as "observed and not found" and N/A as "not observed." This distinction is not a product fact. It is a working rule that helps keep an absent measurement from being confused with a negative result. In the profile, the check is whether the fields follow that rule: a field that is empty because the measurement is unavailable must be N/A, and a zero must correspond to an actual observation. Applied to the question of how AI visibility gets measured, the rule stops N/A and zeros from being added together as if they were the same thing. It is a reading proposal, not a universal standard, and it should be labeled as such whenever profiles are compared.
Where the evidence stops
The limit sits in the record itself. We can state that API responses are not a person's screen and that missing values are N/A, but those facts do not add up to a universal method for measuring AI visibility. Confirming the profile is a consistency check, not a guarantee of results. For the full picture, see the explainer page.