Which AI assistants does Monroya measure?

Paid plans check all 6: ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Google AI Mode. The free plan checks Gemini and Perplexity.

Why ask each question several times?

A single answer from an AI assistant can change minute to minute. We run every tracked question several times per check and look at what the assistant consistently says, not what it said once.

What's the difference between Named, Recommended and Cited?

Named means your business is mentioned in the answer. Recommended means the answer puts you forward as a choice. Cited means one of your pages is linked as a source. We track all three separately because the fix for each is different.

What's a branded question, and why score it separately?

A branded question already includes your business name (for example, "is [your company] good"). We score those separately so your main numbers measure whether AI recommends you to buyers who have not heard of you yet.

How do you detect a possible AI model update?

If 30% or more of an account's tracked questions shift on one assistant in the same scan, or the same shift shows up across many accounts at once, we flag it as a possible model update rather than counting it as your win or loss.

Can I see the actual AI answer behind a result?

Yes. Every answer we collect is saved, so any result can be traced back to the exact AI answer that produced it.

Technical reference

How Monroya actually works.

The companion to /how-it-works for teams who want the detail: repeated checks, how we define Named, Recommended and Cited, what counts as a verified win, and how we catch a possible AI model update.

Updated: October 2026

By Jeremy Unruh, Founder, Monroya · Reviewed October 2026

Assistants checked

Paid plans check all 6 AI assistants below on every scan. The free plan checks Gemini and Perplexity.

Assistant	On paid plans	On the free plan
ChatGPT	Yes	No
Claude	Yes	No
Gemini	Yes	Yes
Perplexity	Yes	Yes
Google AI Overviews	Yes	No
Google AI Mode	Yes	No

Repeated checks

A single question asked once can return a different answer the next time, even on the same assistant. To make a result you can trust, every tracked question is asked several times per check. Your result reflects what an assistant consistently says, not a one-off answer.

Named, Recommended and Cited

Each answer is scored on three separate signals, because each one calls for a different fix:

Named — your business is mentioned in the answer text.
Recommended — the answer puts you forward as a choice the buyer should consider.
Cited — one of your pages is linked as a source behind the answer.

A business can be named without being recommended, and recommended without being cited. We keep the three apart instead of folding them into one score.

What counts as a verified win

A win is not a single good answer. A question counts as a verified win only when the same question, on the same assistant, names you consistently across repeated checks. If a win later slips, you get an alert.

Branded questions, scored separately

Questions that already include your business name (for example, "is [your company] worth it") are tracked but scored on their own. They tell you how AI talks about people who already know you. Your main score is built from questions that do not name you, since those are the ones that show whether AI recommends you to buyers who have never heard of you.

Possible model update detection

AI assistants update their underlying models without notice, and that can shift answers across many businesses at once, independent of anything you did. We flag a scan as a possible model update when either is true:

30% or more of an account's tracked questions shift on one assistant in the same scan, or
the same shift shows up across many accounts at the same time.

Saved answers

Every answer we collect from an assistant is saved. Any result in your account, named, recommended or cited, links back to the exact saved answer that produced it, so nothing is a black box.

What we store

Per check: the saved assistant answers, which questions were named, recommended or cited, and the result history over time. Per account: your tracked questions and competitor setup.

Your scan data is yours. We send your buyer questions to the AI assistants to read their answers, and we don’t sell or share your account data.

Frequently asked questions

Which AI assistants does Monroya measure?: Paid plans check all 6: ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Google AI Mode. The free plan checks Gemini and Perplexity.
Why ask each question several times?: A single answer from an AI assistant can change minute to minute. We run every tracked question several times per check and look at what the assistant consistently says, not what it said once.
What's the difference between Named, Recommended and Cited?: Named means your business is mentioned in the answer. Recommended means the answer puts you forward as a choice. Cited means one of your pages is linked as a source. We track all three separately because the fix for each is different.
How do you define a verified win?: A win counts only when the same question on the same assistant names you consistently across repeated checks, not a single lucky answer.
What's a branded question, and why score it separately?: A branded question already includes your business name (for example, "is [your company] good"). We score those separately so your main numbers measure whether AI recommends you to buyers who have not heard of you yet.
How do you detect a possible AI model update?: If 30% or more of an account's tracked questions shift on one assistant in the same scan, or the same shift shows up across many accounts at once, we flag it as a possible model update rather than counting it as your win or loss.
Can I see the actual AI answer behind a result?: Yes. Every answer we collect is saved, so any result can be traced back to the exact AI answer that produced it.