Every tracked prompt sits in one of five states. The states are not a score; each one points at a different job.
Cited
The answer named you and linked one of your pages. This is the win.
Do: keep the page current. Cited pages get re-read; a stale one loses the position quietly.
Mentioned
It named you, but linked nothing of yours.
Do: earn the citation. The engine already believes you belong in the answer — it just did not have a source it wanted to point at. That is usually a structured-data or sourcing gap, not a content gap.
Cited elsewhere
It answered the question using someone else's page. You were not named.
Do: go where it went. This is the most useful state on the page, because the question has a proven answer and the engine has told you exactly which source it trusts. Getting mentioned on that source is often faster than out-writing it.
Absent
It answered without you at all — no mention, no citation.
Do: write the answer. If competitors were named, the question is contested and you are losing it. If nobody was named, the engine has no confident answer, and that is the cheapest kind of gap to win.
Needs run
Nobody has asked yet.
This is not a loss. It is unknown. A prompt you added this morning has no result because no result exists. Treating "needs run" as absence is how people conclude their AI visibility is terrible on their first day.
How the numbers relate
- Prompts naming you counts prompts where the brand was actually named. A prompt where the engine cited three sources and none were you is not visibility — it is the loss list.
- Sentiment is read from the wording around your name, so it only exists on answers that named you. "Not named" is not a neutral score; it means there is nothing to read.
- Named first is closer to rank one than a plain mention. Being named third in a list of five is not the same result as leading it.
Next
- AI Research vs Prompt Lab vs AI Tracker
- Writing prompts worth tracking