How to Track Brand Mentions in AI Search for Free (and When to Pay)

This page answers the money question. Our full guide to tracking brand mentions in AI search covers the whole system: the three prompt buckets, the recording template, cadence and alert rules. Here you get the free method, a simple test for when to pay, and a price table.

Before you pay anyone to watch brand mentions in AI search, know this: paid trackers don't see more of the engine than you do. Google's AI optimization guide says "No third-party tool has access to our internal ranking or AI systems." Trackers run prompts the way you would, only more often. What you pay for is volume and time.

The Free Method, Step by Step

Free tracking has two parts. You collect answers yourself, because no free report shows brand mentions in AI search across every engine. Then you add the vendor data that Microsoft and Google give away.

A weekly log of brand mentions in AI search

1. Write 15 to 20 buyer prompts. Mix questions about your brand, your category ("best X for Y") and head-to-head comparisons. Keep your name out of the category prompts. The free AI Visibility Prompt Kit drafts 30 to choose from.
2. Pick two or three engines where your buyers ask. ChatGPT, Perplexity and Google AI Mode are the usual start.
3. Use a clean session. Log out, or use a temporary chat without memory, and run from the same place each week.
4. Log six fields per answer: date, engine, prompt, named (yes or no), linked (yes or no), and the other brands and sites cited.
5. Run each prompt once a week and add up four weeks before you read a rate. Weekly swings on a small list are mostly noise.

That's 15 to 20 prompts times two or three engines, or 30 to 60 answers a week. The hub post has a fuller recording template if you want more columns later, and our glossary explains prompt tracking in plain terms.

Three free vendor reports

Here is where each row comes from. Microsoft launched the AI Performance report in public preview on 10 February 2026 and added Citation Share in a June update, noting it "does not expose competitor domains." Google's generative AI performance reports reached every site on 31 August 2026. GA4's channel definition names the five assistants in the table and excludes AI Overviews and AI Mode. Perplexity and Claude aren't named, so add a custom channel with our 15-minute GA4 setup.

Read all three by page once a month. A page that rises in Bing's citations and Google's AI impressions at the same time is one engines trust. Check whether your log shows it too.

Free one-off checkers and trials

Free checkers are good for a first look, not for tracking over time:

• Semrush's AI Visibility Checker covers ChatGPT, Gemini, AI Mode and AI Overviews, and allows three checks a day without an account.
• Ahrefs' free AI visibility checker queries ChatGPT, Gemini, Perplexity, Copilot and AI Overviews with no signup, and calls its report "a limited preview."
• HubSpot's AI Search Grader is a one-time check of what ChatGPT, Perplexity and Gemini say about you "based on their training data," so it reflects what the models learned rather than a live web search.
• Free tiers and trials. AthenaHQ's Essential tier is free with 300 credits, at one AI response per credit. Profound's trial runs 50 prompts a day for 7 days. SE Visible offers a 10-day trial, and Semrush and LLMrefs offer 7 days.

A trial is most useful after a month of manual logging. You'll know which prompts matter, so you can judge whether the tool's numbers match yours.

When to Pay: The Tracking Break-Even Test

We call this the Tracking Break-Even Test. It puts a price on your own time, because free tracking of brand mentions in AI search is only free if your hours are.

Step 1. Price one logged answer. Time your first full session and divide by the answers you logged. As an illustration, say running a prompt, reading the answer and filling the row takes 1.5 minutes, and your time is worth $40 an hour. One answer then costs 1.5 ÷ 60 × $40 = $1.00.

Step 2. Count your monthly answers. Prompts × engines × runs a week × 4.33 weeks.

Step 3. Compare. If answers a month × cost per answer is higher than a plan that covers the same prompts and engines, pay. If it's lower, stay free.

Here is the test for Plannora, a made-up project management tool. It logs 20 prompts in three engines once a week: 20 × 3 × 1 = 60 answers a week, and 60 × 4.33 = 260 answers a month. At $1.00 each, that's $260 of time a month, or 6.5 hours.

Break-even is the plan price divided by $1.00 an answer. Plannora's 260 manual answers are well past every line, so paying is cheaper than its time. But the last column matters more than the price. No plan above matches Plannora's exact engine mix, so it either changes its mix or picks the plan that covers the engines its buyers use. Coverage first, price second.

Signs you should stay free

• You track fewer than about 100 answers a month, so the hours are small.
• You want to read the answers yourself, to learn which sites engines lean on.
• You report to yourself, not to a client or a board.

Signs it's time to pay

• You need more runs than a person can log. Steady numbers on brand mentions in AI search take repeat runs: the 2026 study "Don't Measure Once" advises at least 7 runs per prompt per day for brand visibility.
• You track several countries or languages.
• You need months of history and exports for a monthly report.
• You run an agency and track several brands at once.

Paid Trackers Compared, With Prices From 28 September 2026

These are the cheapest listed ways to have software log brand mentions in AI search for you. Every row comes from the vendor's own pricing page on 28 September 2026, at monthly billing unless noted. For deeper vendor questions, see our ChatGPT rank tracker roundup.

Ahrefs' check rule is "1 prompt execution × 1 LLM × 1 location," per its help center, and the platforms include AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini and Copilot. HubSpot's "week over week" comes from its grader page. Profound's Enterprise plan tracks up to nine engines.

Two listening suites also sell tracking of brand mentions in AI search as an extra. Brand24's pricing page lists an AI Visibility module as an add-on, with plans from $199 a month billed yearly; it runs through Chatbeat, above. Meltwater sells GenAI Lens, typically as an add-on, with no public price.

Price per answer, and why it matters

Prompt counts hide the real unit, which is answers. Three vendors state a monthly answer total, and Otterly's can be worked out:

The gap is large. Even the most expensive paid plan here costs under 4 cents an answer, against about $1 of your time. That's why the test above tips toward paying fast once your list grows. It's also why free remains the right start: you learn which prompts deserve those cheap answers.

What Rankbox Does After You Find a Gap

Rankbox doesn't track brand mentions in AI search or citations today, so it isn't in the table. It helps with what comes next. Answer-Space Research maps the questions buyers ask AI in your category, which also makes a good prompt list. The Citation-Ready Writer then writes source-backed articles for the prompts where rivals are named and you aren't, delivered to your site through Rankbox's API. The Business plan is $49.50 a month with a 7-day trial; see pricing.