no_surface
Updated 2026-09-15
no_surface is the status recorded when the AI surface being measured did not appear at all for a query — most often an AI Overview that Google chose not to generate. It is neither a mention nor a miss, and collapsing it into either one corrupts the metric.
How it changes the arithmetic
Rates are computed over answers that exist. A query that produced no surface contributes nothing to the numerator and nothing to the denominator; it is reported separately as a count so the reader can see how often the surface was available at all. The alternative — treating it as a zero — makes your number worse whenever Google becomes more conservative, which is a change in Google's behaviour rather than in your visibility.
Illustrative cycle (example numbers)
Ten queries checked three times: 18 checks produced an overview, 12 did not. The brand appeared in 5 of the 18. The reported rate is 5/18 ≈ 0.28, with "no surface in 12 of 30 checks" stated next to it — two facts, neither hidden inside the other.
Common mistakes
- Averaging no_surface checks in as misses, the most common way these numbers are quietly wrong.
- Dropping the count entirely, which hides that the surface is disappearing for your category.
- Confusing no_surface with provider_error — one is the engine's decision, the other is a failed request.
Frequently asked questions
- Should no_surface queries stay in the prompt set?
- Yes. The trend in how often the surface appears is itself a finding, and removing those queries would hide a structural change in the category.
- Does no_surface mean the query is unimportant?
- No — it means the engine decided a generated summary was not appropriate for it that time. The same query can produce an overview a week later.
