How to Track Your Brand in AI Answers Without a Paid Tool

You can track how AI tools talk about your brand with a spreadsheet and thirty minutes a week. Write 20 buyer prompts, run them in each engine, and log whether you are mentioned, linked and described correctly.

Paid AI visibility tools exist and can be useful later. But the method behind all of them is the same, and a small team can run it by hand and learn more from reading the answers directly. This guide shows the setup, the protocol and how to read the results. It pairs well with AI search optimization best practices, which covers what to change once you see the gaps.

What should you measure?

A common surprise is a page that is linked as a source while the brand is not named in the text. Logging mention and citation separately shows that gap.

How do you build a 20-prompt set?

Write prompts the way your buyers would type them, in four groups of about five:

• Category: “What are the best [category] tools for [type of company]?”
• Alternatives: “What are good alternatives to [competitor]?”
• How-to: “How do I [job your product does]?”
• Brand: “What is [your brand]?” and “Is [your brand] good for [use case]?”

Take the wording from sales calls and support tickets where you can. Freeze the list once you start, so week-to-week comparisons are fair, and add new prompts in a separate tab.

What is the weekly run protocol?

1. Open a clean session in each engine you track: ChatGPT, Gemini and Perplexity at minimum, plus Google AI Overviews and Claude if your buyers use them. Log out of personalization where you can.
2. Paste each prompt exactly as written.
3. Record the result for the five measures and paste the link list.
4. Note the date and the engine.
5. Once a month, repeat a few prompts several times to see how much the answers vary.

Why can a single run mislead you?

AI answers are generated fresh each time. Brand lists change from run to run, and studies of AI citations cited by Advanced Web Ranking report that the sources behind an answer can turn over quickly from week to week. A single result tells you very little. What is more stable is the set of brands a tool tends to consider for a topic, and your presence across many prompts over several weeks. Judge by the trend: in how many of the 20 prompts were you mentioned this month compared with last month?

A spreadsheet layout you can copy

Add a second tab that counts, per engine and per week, the share of prompts where you were mentioned. A simple line chart of that number is the report.

Which other data sources should you check?

• Search Console: Google says it reports on its generative AI features. Check which pages and queries appear.
• Analytics: in GA4, build a segment for referral sources such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com, and watch sessions and conversions.
• The source lists in the answers themselves: note which sites are cited for your category. Those are the places to earn mentions.

What does a good first month look like?

Week one gives you a baseline, and you will probably be named in only a few of the 20 prompts. Weeks two to four show which of your changes matter. A reasonable first-month goal is to know exactly which prompts you appear in, which sources the answers cite and which two pages or placements would most likely change the picture. Visibility itself tends to rise afterward.

What do you do with the results?

• If you are missing from category prompts, strengthen the pages that answer them and earn mentions on the sites the answers cite.
• If you are mentioned but described incorrectly, fix the facts on your own pages, then on your profiles, and use the same wording everywhere.
• If you are cited without being named, add your brand and product name near the key claims on that page.
• If a competitor always appears first, read the page the answer cites and see what it covers that yours does not.

When is a paid tool worth it?

When you need more than about 50 prompts, more engines, daily runs or team dashboards. The spreadsheet method will have taught you what to look for, which makes any tool easier to judge. Whatever you use, ask how it handles repeated runs, because a single snapshot hides most of the variation.

A 30-minute weekly routine

1. Run the 20 prompts across your engines (20 minutes).
2. Log the results and update the chart (5 minutes).
3. Write down one action for the week, such as fixing one page or pitching one site (5 minutes).

Want a baseline without building it yourself? Our free audit runs a starter prompt set across the major engines and shows where you stand. For the plan around it, see our AEO strategy guide.