AI VISIBILITY MEASUREMENT AND RECOMMENDATION PLATFORM
See why AI recommends your competitors and not you.
Metrisque measures how ChatGPT, Gemini and Claude understand your brand, products and pages. It shows each gap and the change to make to get recommended.
Diagnose measures how ChatGPT, Gemini and Claude read your company.
WE MEASURE
The assistants we measure.
READS THE SAME CATALOGUE
Agents that discover products through Shopify Catalog and the Universal Commerce Protocol. The same product data we measure.
BUYER QUESTION
SAMPLE · CONSUMER PRODUCT
“Best waterproof hiking boots for wide feet”
Competitor A
82%
Competitor B
64%
Competitor C
51%
Your brand
18%
Unit: percent of recommendation slots won on this question, 0 to 100 percent. Models: ChatGPT, Gemini and Claude, 3 runs each. Question: “Best waterproof hiking boots for wide feet”. Read on 14 July 2026. Sample result, brand anonymised.
ChatGPT
Gemini
Claude
LARGEST GAP
Your product pages never state width fitting. The three shortlisted brands do, on the page the models read.
AFTER THE FIX, REMEASURED ON 11 AUGUST 2026 · PERCENT OF SLOTS
18%→47%
THE INTRODUCTION
1 MIN 13 SEC
What Metrisque does, in just over a minute.
WHAT METRISQUE IS
An AI search visibility platform for businesses that sell online.
It tracks whether ChatGPT, Gemini and Claude recommend you on real buying questions, explains why, and re-measures after you publish a fix.
CHATGPTLIVE ANSWER
BUYER ASKS
“Best CRM for a small design studio”
THE ASSISTANT ANSWERS
You were not named.
FROM OUR PRE-REGISTERED STUDY
Rankings barely explained who got recommended. Understanding did.
92%
of AI recommendations came from inside the measured territory
A recommendation is the end of a path. It is the end of the buyer's journey. Every step before it can be measured.
An assistant has to know you, understand you, find you relevant, back it with sources, and shortlist you. We measure every step, tell you whether the question is worth fighting for, and which words to change to move it.
No equivalent to the fix. Counting cannot change the answer.
Both sides describe the same question. One is measured before the answer exists, and comes with the fix. The other is counted after, and can only be watched.
WHAT YOU GET BACK
Three plain answers on every question your buyers ask.
Predict
Where do you stand today?
Your position on each question, beside the competitors named instead of you.
Attribute
Why does it read that way?
The page text, the missing facts, the sources. No opaque score.
Verify
Did the change work?
The same questions run again, compared against your stored baseline.
HOW IT WORKS
You get from guessing to a published fix in four steps.
1Measure
Enter your website. We read your public pages and run your market's questions across the assistants.
2Find
See which step is costing you, and who holds the slot you are missing.
3Fix
Get the change to make, the evidence it needs, and a rewrite in your own wording.
4Verify
Publish, re-measure, read the movement against your earlier run.
The buying questions measured, in their own words.
Who was recommended instead, per assistant.
The change that was made, and on which page.
The re-measurement, with both run dates.
QUESTIONS
AI search visibility, answered
What is AI search visibility?+
How often AI assistants name your business when buyers ask what to choose. Not a search ranking: the assistant builds its list from what it recalls about you, how it classifies your pages, and which sources it can cite.
How do I check my AI search visibility?+
Enter your website. We run your market's buying questions across ChatGPT, Gemini and Claude and return your position on each, who was named instead, and the page text behind it. Under a minute.
What does Metrisque measure?+
Six measures and a gate before them: whether assistants recall you, read you as the right kind of business, match the buyer in the question, name you in the picks, and which sources back the answer.
How is this different from SEO tools?+
SEO tools report search rankings. We report AI recommendations, a different list. In our study the two overlapped weakly and the order did not match.
Which AI models do you measure?+
ChatGPT, Gemini and Claude, each on recall, live search and recommendation questions. Category Fit is read twice, by two independent rulers, so you can see where they disagree.
Can Metrisque guarantee AI will recommend me?+
No. Answers vary between identical runs, so nobody can promise selection. We measure the parts you control: recall, category, buyer fit and evidence. In our study, 92 percent of recommendations came from inside that territory.
Which businesses can use it?+
Ecommerce brands, software companies, B2B service firms, and agencies running it across client portfolios.
How long does a measurement take?+
Under a minute for the first read. Deeper runs cover more questions and pages, and land in your reports when they finish.
Do you change my website?+
No. We write nothing to your site. You get the change to make and a rewrite to copy. Where a push integration exists, you approve every change first.
How do you verify a fix worked?+
By re-measuring. The same questions on the same pages after your change, compared against the stored run, with both dates beside it.
Find out what AI says about you.
See where you stand on the questions your buyers already ask. Free, about a minute.
