Skip to main content
Visibility tells you how often the AI names your brand. Intent tells you in which kinds of question it names you, what the AI did with those questions, and how your brand was named. “Named” covers everything from “my top pick” to seventh in a list of nine, and a mention in an explanation is not the same as a mention in an answer that sends the reader to a shop. Bluemoon labels intent at three levels: Questions and answers use the same three labels, so you can compare them directly: a commercial question that gets an informational answer means the AI explained instead of comparing. The Intent map shows that comparison across all your answers.

​
Where to find it

Intent map

Intent map in the left sidebar: what was asked against how the AI answered, your share per question type and per platform, and how close your questions sit to a purchase.

Prompts

The Intent filter on the Prompts list narrows your questions by intent. A question’s page shows what each platform did with that question.

Every AI answer

Open any answer. The viewer shows the question’s intent, the answer treatment and the evidence behind it, the role of each tracked brand named, and the answer’s stage.

​
Question intent

Every tracked question gets a commercial intent score from 0 to 100: how close the person asking is to a purchase. The higher the score, the closer to buying.
  • On the Intent map, scores are grouped into four bands: Ready to buy (75–100), Comparing (50–74), Exploring (25–49) and Just curious (0–24).
  • Next to answers, the same score is shown as one of three labels (Informational, Commercial, Transactional) so it lines up with the answer treatment.
  • New questions are scored automatically shortly after you add them. Until then they show as Not scored: they are left out of the Intent map and counted in a note under it, never folded into Informational.
  • The score changes when the question changes, so it stays stable from one week to the next.
The score helps you prioritise. It is not a prediction. Use it to decide which questions deserve your attention first, not as a forecast of where the AI will name you.
If Settings → Domains & markets shows No market set for a domain, add a country. Without one, the intent of that domain’s questions is estimated from their wording alone, which is less precise.

​
Question types

Each question also has one type: Definition, How to, Comparison, Best-of list, Alternatives, Pricing, Review, Recommendation, Troubleshooting or Availability. The Intent map shows how often the AI names you for each type, with the number of questions and answers behind each figure.
  • A type appears once its questions have been asked enough times. The empty panel tells you how many.
  • A figure based on too few answers shows a dash instead of a percentage.
  • A type is flagged Blind spot only when your rate there is clearly low and enough of your questions have that type. If no type qualifies, nothing is flagged.

​
Answer treatment

The treatment is what the AI did with the question. An answer can be both transactional and commercial. The Intent map counts it under the label closest to a purchase; the answer viewer lists both. In the answer viewer, the treatment comes with the evidence behind it, so you can check the label against the answer itself. Some labels are marked refined by an AI read of the answer: they were settled by reading the text of the answer.

​
Brand role

The role says how one brand was named inside one answer. Every answer that names at least one tracked brand gets a role for each of them: yours and each competitor’s. The answer viewer shows what each role was based on, so you can check it against the text.
Roles only take into account the brands you track. “First” means first among your tracked brands, not first in the answer.

​
Stage

In the answer viewer, each fully classified answer also shows a Stage: An answer whose roles have not been read yet shows no stage.

​
”Not measured” is never zero

A value Bluemoon has not measured is shown as a dash or as not measured yet, never as 0 or 0%.
  • An answer whose roles have not been read yet shows Roles not read yet.
  • “Recommended in X of N answers that named you” appears only once at least one of those answers has had its roles read. “Recommended in 0 of 41” is a finding; “not measured yet” means nobody has looked yet.
  • A question with no intent score is left out of the Intent map and counted, not hidden.
  • A measured rate that rounds to 0.0% is shown as <0.1%: a handful of mentions is not the same as none.

​
When the labels appear

After a scan, each answer’s treatment shows right away. Brand roles and AI-refined labels follow a few minutes later; until then, the screens say so. On long windows, older answers may have no roles: they show a dash on role figures, never a zero.

​
Limits

An answer reads as transactional partly because of the shops it cites. Early in an account’s life, few cited sites have been recognised as shops, so fewer answers read as transactional. The Intent map tells you how many shops have been recognised so far.
A low share of recommendations tells you the AI did not pick you when it named you. It does not tell you why, and Bluemoon does not claim to know. Open the answers and read them: the evidence is one click away.