AI mention tracker, with the answers behind every number
See how often eight AI engines name your brand when buyers ask your questions, who they name instead, and whether a change is real or run-to-run noise.
- All eight AI engines on every paid prompt run
- A 95% interval on visibility and citation rates
- Every run priced before it starts
Opening soon: sign-up opens with live measurement, after our provider pilot.
238 answers to 8 prompts over 4 weekly runs. Google showed an AI Overview for 14 of 32 searches.
ChatGPT47%
Gemini41%
Perplexity56%
AI Overviews57%
AI Mode47%
Claude53%
Grok38%
Muse47%
RookAtlas is an AI mention tracker. It asks ChatGPT, Gemini, Perplexity, Google AI Overviews, Google AI Mode, Claude, Grok and Muse the questions your buyers ask, on a schedule, and records which brands each answer names: the first five as a logged-out visitor sees them, Claude, Grok and Muse through their makers' official APIs. Your visibility is the share of answers that name you, pooled across the engines with a 95% interval, and every engine gets its own figure beside it.
Every prompt, on every engine
A grid of your prompts against the eight engines shows where you're named in every run, where you're named some of the time and where you're missing. Open a cell to read the answer itself.
- Each cell counts the runs that named you
- The latest answer with every brand it named, in order
- The sources that answer showed
| Prompt | ||||||||
|---|---|---|---|---|---|---|---|---|
| P-01Best noise cancelling headphones for a long-haul flight | ||||||||
| P-02Best wireless earbuds for running | ||||||||
| P-03Best earbuds to use with an iPhone | ||||||||
| P-04Cheaper alternatives to AirPods Pro | ||||||||
| P-05Are JBL headphones good value? | ||||||||
| P-06Best headphones with spatial audio | ||||||||
| P-07Best headphones for calls when working from home | ||||||||
| P-08Best headphones for a daily commute |
alwayssometimesneverno answers
JBL named in 113 of 238 answersChanges you can tell from noise
AI answers vary from run to run, so a rate without a range claims too much. Visibility and the cited rate carry a 95% interval, and a change is tested on the prompts asked on the same engines in both periods, each counted once: from 10 such prompts RookAtlas tells it apart from run-to-run noise, and calls it a change only if the next period agrees.
- Intervals that account for prompts clustering
- Too little data says so instead of drawing a bar
- The same estimator in the app, reports, API and MCP
Named, cited, or both
Being named and being cited are different problems with different fixes. RookAtlas tracks both on every answer, so you know whether to win the mention, the source, or neither.
- Mentions in the answer text, with or without a link
- Citations of your own pages, with or without your name
- The action that fits each case
How it works
- Step 1
Add your brand
Your site, the names people use for you, and the competitors you lose deals to.
- Step 2
Choose the questions
Write the prompts buyers ask, or start from suggestions. Each prompt is versioned, so an edit starts a new series.
- Step 3
Run on eight engines
On a schedule or on demand. Every run shows its credit price before it starts.
- Step 4
Read the answers
Open any answer to see who was named, in what order, and which pages it cited.
What's inside
Mention rate with intervals
The share of answers naming you, with a 95% interval and the number of answers behind it.
Every engine apart
One pooled number, and each engine's own rate beside it, picked in any mix.
Recommendation labels
Planned: distinguish endorsements, alternatives and cautions after the extractor evaluation.
Markets
A country and a language per market, so you see the answers people there get.
AI Overviews trigger rate
How often Google shows an Overview at all, kept apart from how often it names you.
Versioned prompts
Editing a prompt starts a new version, so a trend never mixes two questions.
What it can't do
RookAtlas sees what ChatGPT, Gemini, Perplexity and Google's AI Overviews and AI Mode answer a logged-out visitor in the market you choose, and what Claude, Grok and Muse answer through their official APIs, which can differ from their apps. Signed-in answers shaped by a person's memory and history can differ too. The interval covers sampling within your prompts, not whether those prompts represent every buyer, so choose them with care.
Works with
All featuresQuestions
Prices, credits and plans are on the pricing page .
What counts as a mention?
An answer that names your brand by its name or an alias you've added. An answer counts once however often it repeats the name. We record its first-appearance rank. When sentiment analysis is enabled, answer details also show list positions and recommendation labels.
Why does my visibility change between runs?
AI answers vary even for the same question on the same day. The 95% interval shows how much of a move could be that variation. RookAtlas compares the prompts asked on the same engines in both periods and, from 10 such prompts, tests whether the change is more than that variation; it counts as a change only if the next period agrees.
How many prompts do I need?
Enough for the interval to answer your question, and at least 10: a rate from fewer prompts stays low confidence however many answers it has. More prompts narrow the interval faster than more runs of the same prompts, because answers to one prompt resemble each other. Visibility's interval shows when you have enough.
Does it count mentions without a link?
Yes. A mention is your name in the answer text, linked or not. Citations of your pages are tracked separately, so you can see where you're named but not cited, and the reverse.