Answers
Brand missing from an answer: what to check
When your brand doesn't show up in an answer, the first step is to work out which kind of absence you're looking at. RankAI separates three cases: a detected absence, an unavailable measurement, and coverage that doesn't apply.
When your brand doesn't appear in an answer, the first thing to check isn't an explanation. It's which kind of absence you're looking at. RankAI separates three cases: detected absence (the answer exists, it was observed, and the brand isn't in it), unavailable measurement (the answer or the metric wasn't collected, and the product shows it as N/A, never as zero), and coverage that doesn't apply (that engine or language is outside what RankAI measures). Mixing up these three leads to bad diagnoses: an N/A read as zero sets off an alarm, and a detected absence treated as missing data leads to waiting for nothing.
Unlike the articles next to it, this one doesn't restate their arguments. V03 and V04 already covered the difference between an API response and a person's screen, and between Gemini and Google AI Overviews as separate observations: here those points serve as proposed checks, not as new facts about the product. V06 already covered N/A versus zero: here that topic is reworked into a three-state diagnosis. What we add is a working artifact — the answer card — and a way to classify each absence into one of the three states.
The three checks, in order
Work through them in this sequence:
- What kind of source is it? Is the answer you're looking at from a human screen, an API call, Gemini, or Google AI Overviews? These are different observations and shouldn't be added together. If you're looking for the brand in an answer on ChatGPT or Google, note which of these sources you have in front of you before drawing any conclusion.
- Is the measurement there? Does the metric exist and was it collected? If not, RankAI shows it as N/A. An N/A is not a zero: it means the measurement is not available, not that the brand is absent.
- Coverage. Is that engine or language inside what RankAI measures? A page in simplified Chinese, for example, does not mean RankAI measures 豆包, 元宝, Kimi or 千问. If coverage doesn't apply, the absence tells you nothing about the brand.
The answer card: a way to keep the three states apart
As a RankAI working method, not a universal standard, we suggest filling in a minimal card for every answer you observe, with five fields: source (screen, API, Gemini, AI Overviews), query, date of observation, measurement status (available or N/A), and coverage (applies or doesn't apply). Only when the first four fields are filled in and coverage applies can you call it a detected absence. In every other case, the card itself shows why you can't conclude anything.
How each state shows up
Detected absence: the card is complete, the measurement is available, coverage applies, and the brand isn't in the observed answer. This is the only case where the absence tells you something about the brand.
Unavailable measurement: the card shows N/A where the value should be. It's not zero and it's not an absence: it's a gap in collection. Treating it as an absence means diagnosing a problem that was never measured.
Coverage that doesn't apply: the card shows that the engine or language is outside RankAI's coverage. The absence says nothing about the brand, because there was nothing to measure in that scope.
What to do after classifying
If the card says detected absence, the next check is to repeat the observation on the same source and the same query, to see whether the absence is steady or occasional. If it says unavailable measurement, the check is on collection, not on the brand. If it says coverage that doesn't apply, the check is on scope: that question belongs to another observation, not to RankAI. In none of the three cases do we promise the brand will show up: we don't know whether or when it will, and this article doesn't claim it.
During pre-launch, checkout can stay off. The invitation is to create an account or open the relevant page, not an immediate trial.
What this article adds. The three-state working distinction — detected absence, unavailable measurement, coverage that doesn't apply — and the answer card as the artifact that keeps them apart. It isn't a universal standard: it's a RankAI proposal to stop an N/A being read as zero, or an absence outside coverage being treated as data about the brand.
Limits. This article doesn't establish why a brand is missing, and it doesn't predict whether it will appear: it only describes how to classify what you observe. The points about API versus screen and about Gemini versus AI Overviews refer to observations already published (V04, V06) and serve here as checks, not as new facts. For the full picture, see the in-depth page.
Why an N/A is not a zero: how RankAI handles an unavailable measurement