Pick 15 to 30 fixed customer questions, ask each one in fresh, logged-out sessions at least twice per engine, and record who’s named, in what order and which links are cited. Track the mention rate over time, not single answers. You can do this in a spreadsheet, with a self-serve tool ($29 to $500+ a month), or with a done-for-you monitor.
Why tracking AI mentions is tricky
- Answers vary per run. The same question can name different vendors, in a different order, a minute later.
- Logged-in vs logged-out. Chat history and memory can personalize answers.
- Location. Answers change by country and sometimes by city.
- Models change. A new model version can shift answers overnight.
- Google’s AI Overviews don’t always appear. Track whether one showed, not just whether you were in it.
So track a fixed set of questions, repeat each one, and look at rates over time rather than single answers.
Option 1: The manual spreadsheet method
- Write 15 to 30 customer questions you’ll keep fixed: “best [service or product] for [audience]”, “best [service] in [city]”, “[competitor] alternatives”, “[category] pricing”.
- Set up columns: date, engine, question, run number, named (yes/no), position, competitors named, cited URLs, notes on accuracy and tone.
- Ask each question in a fresh, logged-out session, twice per engine.
- Calculate your mention rate per engine (answers naming you ÷ total answers) and your average position.
- Repeat monthly with the same questions.
Pros: free, and you see the actual answers. Cons: slow, easy to drift in method, and hard to scale past a few dozen questions.
Option 2: Self-serve tracking tools
A growing number of AI search monitoring tools run your prompts on a schedule and chart the results. Pricing commonly runs from about $29 to $500+ a month (public pricing pages, checked November 2026; prices change), depending on prompt volume and engines covered.
Pros: automated, scalable, good charts. Cons: you still interpret the data and decide what to fix, and some tools sample only a few runs or a subset of engines. Check how each one collects answers.
Option 3: Done-for-you monitoring
A service captures the answers for you and tells you what changed and what to do about it. Our Monthly Monitor is a done-for-you ChatGPT and Google AI brand monitoring service: $290/month for 25 questions captured weekly across ChatGPT, Gemini, Google AI Mode and Google AI Overviews, with a monthly report and 5 prioritized actions; Growth is $590/month with 60 questions and monthly page drafts. Month-to-month, invoiced.
Pros: analysis and next steps included. Cons: costs more than a basic tool, and you’re trusting someone else’s method, so ask to see every captured answer (we include a CSV).
Metrics that matter
| Metric | What it tells you |
|---|---|
| Mention rate | Share of answers that name you, per engine. |
| Average position | Where you appear in the list when you’re named. |
| Share of voice | Your mentions vs competitors’ on the same questions. |
| Citation rate | How often your own pages are cited as a source. |
| Cited sources | Which third-party sites engines rely on in your category. |
| Accuracy | Whether your category, pricing and audience are described correctly. |
In our sample audit, the sample brand’s overall mention rate was 92%, but it varied by engine: 100% in ChatGPT and Gemini, 83% in Google AI Mode and AI Overviews. Its own site was cited in 8 of 26 answers. One headline number would have hidden both.
Sample data: our anonymized sample audit of a well-known sales CRM (not a client), November 2026, 26 answers across 5 engines. Results vary from run to run.
Frequently asked questions
How often should we check?
Monthly is enough for most small businesses; weekly captures smooth out the noise better. What matters is a consistent question set and method, so changes reflect the engines, not your process.
Why log out to check?
Logged-in ChatGPT can use your chat history and memory, which skews answers toward what you’ve asked before. A logged-out or fresh session is closer to what a new customer sees.
What is LLM visibility tracking?
Another name for the same thing: measuring how often large-language-model tools such as ChatGPT and Gemini mention and cite your brand.
What does done-for-you monitoring cost?
Our Monthly Monitor is $290/month for 25 questions captured weekly across 4 engines, or $590/month for 60 questions plus monthly page drafts. Month-to-month. See the details.
The $750 AI Visibility Audit runs 30 of your customers’ questions across 4 engines, twice each, against up to 4 competitors.