Method
A Proposed Query Is a Phrasing, Not a Search Volume
A query suggested by the model is a way of framing content, not a demand estimate. RankAI's scope does not include volume calculation, and a missing visibility figure shows as N/A, never as zero.
No: a query proposed by the model is a phrasing, not a search volume. If you are looking at a card with ten suggested questions for an article, those questions describe how the content could be framed, not how many people type them. Within RankAI's scope, volume calculation is not part of what the product does: the editorial proposal and the quantitative figure are two separate objects, and they should be read at two different moments. The reference search phrase is there to orient the work, not to estimate demand. If you need a number on AI visibility, you do not get it from the proposed query: you look for it in the measurement, and if the measurement is not available, you find it as N/A.
Why the distinction matters in practice
People working on content tend to treat every phrase as a potential data point: if it appears in a list, it looks as if it has already been measured. But a proposed query comes from an editorial operation — choosing an angle, covering a variant, covering a doubt — not from a count. Confusing the two levels produces two symmetrical errors: useful phrasings get discarded because they "have no volume," or pages get planned around phrases nobody has ever measured. Keeping the two objects separate avoids both.
How to read a card with proposed queries
RankAI's working proposal, to be understood as a method of work and not as a universal standard, is this:
- Read the query as a phrasing. Ask yourself which angle it opens and which part of the answer it covers, not how "big" it is.
- Look for the measurement in a separate field. AI visibility is measured through the relevant observation; the reference search phrase is not that measurement.
- If the measurement is missing, accept N/A. In the product, an unavailable visibility figure is shown as N/A, never as zero: zero is an observed value, N/A is an absence of observation.
- Do not turn absence into an estimate. N/A does not authorize guessing a number, high or low.
The case of an external volume
You may have a volume figure produced elsewhere. If it exists but its source is not declared, or it is not comparable with the scope in use, mark it as "source not declared" and do not use it for editorial decisions. A number without provenance is not data: it is an opinion with digits.
What this distinction adds beyond a simple filled-in card is the direction of the check. Faced with a proposed query, the right question is not "how much is it worth?" but "what am I looking at?" If the answer is "a phrasing," the next step is to look for the measurement elsewhere; if the answer is "a measurement," the next step is to verify its object and its provenance. Running this sequence in reverse — starting from the number and looking for the phrase that justifies it — is the fastest way to give an editorial proposal a weight it does not have. The separation, then, is not a formality: it is what keeps a content choice from turning into an implicit data point. And when the data is missing, the product says so with N/A instead of filling the gap with a zero, which would itself be an unobserved claim.
A verifiable limit
The limit lies in the scope: within RankAI's scope, volume calculation is not part of what the product does, so no proposed query should be read as a demand estimate. What comes back from an engine through an API is the API response, not the screen a person sees; and Gemini and Google AI Overviews are not the same observation. That is why AI visibility has to be measured with the observation that corresponds to it, not inferred from the phrasing.
For the full picture of how the product works and what falls within its scope, see the in-depth page.