Category misplacement shows up as two symptoms at once: you are absent from the queries you care about, and you turn up in queries you do not. Neither looks like a problem on its own, which is why it goes undiagnosed for so long.
The check takes three prompts: ask an engine to define your category and list who is in it, ask what your product is, ask what problem you solve. Where those answers disagree is the finding.
This is the diagnosis people reach last, usually after a year spent fixing a visibility problem that was never one. The engine knows who you are. It has filed you where your buyers are not looking.
What does misplacement actually look like from the outside?
Like two unrelated problems, which is the difficulty. Teams notice one symptom and never connect it to the other.
Ask for the best tools in your category and you are not listed, even though you are well known enough that a branded question returns an accurate description of you.
You appear in answers about an adjacent, usually broader category, alongside products your sales team has never once competed against.
The branded control separates this from ordinary low visibility. If asking about you by name returns a fluent, accurate paragraph, the engine is not failing to retrieve you. It is retrieving you for the wrong question.
Why does a model have a category for you at all?
Because a category answer has to be supported by sentences somebody wrote. Anthropic’s citations documentation describes the feature as returning “the exact passages that support each claim, so you can verify answers and surface sources.”
For your name to appear in “the leading tools for X are A, B and C,” some retrievable passage has to place you in X. Not imply it. State it.
If no sentence anywhere says you are an X, no answer can say it either. Category membership is an editorial artefact before it is a retrieval outcome.
Which is why this concentrates in companies with the strongest positioning. A team that has defined a new category has, by design, removed itself from the existing one, and the existing one is where the demand lives.
How do you actually run the check?
Three prompts per engine, in this order, because the later ones would prime the earlier ones.
Run all three on every engine, unchanged. Google notes in its documentation on AI features that “AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary.” A misplacement on one engine is a different problem from a misplacement everywhere.
How do you read the result?
Four outcomes from the first two prompts, and only one is a placement problem.
| In the category list? | Self-description matches? | What you have |
|---|---|---|
| Yes | Yes | Correct placement. Any remaining problem is rank within the category, not membership. |
| No | Yes | An awareness problem. The model agrees you are an X; too few sources say so. |
| No | No | Misplacement. The model holds a confident, different category for you. |
| Yes | No | Unstable placement. You sit on a boundary, worth watching rather than fixing. |
The third row is the expensive one and the second is the common one. Telling them apart takes twenty minutes and saves a year of writing for a problem you do not have, which is the broader point in why is my brand not showing up in ai search.
What put you in the wrong category?
Usually one of three things, ranked here by how often they turn out to be the cause.
Your own pages never say the category noun. They say what you do, in your language, with a coined term in the headline. That reads well to someone who already understands the space and gives retrieval nothing to file.
Third parties describe you differently, and there are more of them than there are of you. A directory listing, a roundup and a few forum threads using the same wrong noun outweigh your homepage. The community half of that is in reddit ai search citations brand.
Or you changed category and the record did not. That is the most fixable version, because the right sentences already exist on your site and the stale ones are elsewhere.
How do you correct it?
By making the plain sentence exist in several places, then waiting. There is no technical shortcut.
Start on your own pages. In the first hundred words, put the category noun your buyers use in an ordinary declarative sentence next to your name. Keep the coined term; it just cannot be the only thing you say. Google’s documentation on AI features states that “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary,” so this is editorial rather than technical.
Then correct the third-party record, where the weight actually is: the directories, roundups and comparison pages that describe you. Finding and prioritising them is covered in third party sources ai citations audit. Comparison pages count double, because naming you beside a competitor asserts category membership implicitly, which is the mechanism in how does ai answer x vs y questions.
What does not work is arguing the category into existence on your blog. Every new page about your coined term reinforces the filing you are trying to change.
How do you know it moved?
Re-run the same three prompts, unchanged, and watch two things: whether your name enters the category list, and whether the noun in the second answer changes. The second usually moves first.
Microsoft’s AI Performance report in Bing Webmaster Tools gives you a free, first-party read on the same question, since it shows “the key phrases the AI used when retrieving content” from your site. If the phrases retrieving your pages are all in the wrong category’s vocabulary, that is the misplacement visible from the other direction.
Expect quarters rather than weeks, and set the cadence accordingly instead of checking daily and reading noise, which is the arithmetic in how often should you track ai visibility. Run the peer set from prompt three alongside it, because a wrong category and a wrong competitor list usually share a cause, diagnosed in ai comparing us to wrong competitors.