Skip to content
Sealmetrics

Free · no signup · live

What do AIs say about…?

A company, a brand or a team. Not a person: this is about organisations.

If someone asked about this brand in the past week, the result appears at once. If not, answers appear as each model replies.

How it works

Two questions,
every model at once.

01

Two questions

What the brand is, and whether the model would recommend it to someone about to buy. The same two questions go to every model at the same time, and almost all of them answer from memory, not from what they find that day. The exception is Perplexity, which searches the web before it answers and cites where it looked: it is in the panel because that is what people who ask it actually see.

02

A judge that is not a person

Another model reads each pair of answers and files it as correct, partial, confused with someone else, or no answer. Nobody checks it by hand, and every result says so underneath.

03

One result per brand

Each result is public and has its own address, so you can link to it. If someone asked about the same brand in the last seven days you see that run, and after thirty days it is deleted.

What it is not

  • Not what ChatGPT answers with browsing on. That is a live search result, not the model's memory, and it moves for other reasons.
  • Not a score. Every figure says how many models it comes from, because a number without its denominator is one nobody can check.
  • Not a ranking. It counts, in one run, how many models know the brand, confuse it or recommend it, and says exactly that.
  • Not the full report either. This asks two questions; the AI brand monitoring report asks six, sets you against your competitors and arrives by email.

Two questions give you a snapshot.
Six give you the report.

The full report also asks what the models praise and criticise, how they read your sector, who they put in your place and how they compare you with a competitor. Every answer arrives whole, with the corrections worth making at the source.