How to track brand mentions in AI answers, step by step

Citable · 1 November 2026 · 9 min read

To track brand mentions in AI answers, run a fixed set of category, comparison and recommendation prompts in fresh sessions. Repeat each prompt several times and log whether your brand is mentioned, recommended or cited. Report the rate across runs, because AI answers change from one run to the next.

This guide covers the manual method first, then where free reports and paid tools fit.

What do you need before you start?

You need four things, and none of them costs money:

A spreadsheet: one row per prompt run, so you can calculate rates later.
A competitor list: three to five brands you expect to be named next to yours.
A clean browser setup: a fresh conversation for every run. Turn off personalization or memory where you can, or note that they were on, per Ambeltek's guide.
A few hours: the first pass is slow. Later rounds go faster once the prompt set is fixed.

How do you track brand mentions in AI answers, step by step?

Follow these in order. Each step says what to do, how to know it worked and the usual mistake.

Write a fixed prompt set

Cover category, comparison, problem and recommendation intents. Guides disagree on size. Arahi suggests 20 to 30 buyer prompts across four types, while Omnia suggests 10 to 20 high-intent prompts as a baseline. Ambeltek recommends 30 to 50 for most brands and says fewer than twenty is too small a sample to show a trend.

It worked if every prompt sounds like something a buyer would type. The typical mistake is filling the set with "tell me about [brand]". Agility Writer points out that these prompts reveal little, because a buyer who already knows your name would have found you anyway. Recommendation-style prompts are the more useful test. Citable's guide to checking if ChatGPT recommends your brand covers the same idea in more detail.

Run each prompt in a clean session

Open a fresh conversation for each run. Record whether web search was active, and turn off memory where possible, as Ambeltek advises.

It worked if two runs of the same prompt don't share context. The typical mistake is asking prompts back to back in one long chat, so earlier answers shape later ones.

Repeat every prompt

A single run tells you little. According to Arahi:

"Three runs is the floor for a manual check; five is better for your most important prompts." — Arahi's guide

The reason is variance. In a SparkToro study with Gumshoe.ai, ChatGPT and Google's AI in Search returned the same brand list less than 1% of the time across repeated runs of the same prompt. Treat that study with care: PPC Land notes it was not peer-reviewed and was done with a partner whose business is visibility tracking.

It worked if each prompt has at least three logged runs. The typical mistake is keeping the one run where your brand showed up.

Log what happened

Record the same fields every time. Ambeltek suggests a log that captures:

Whether the brand appeared.
Whether a source was cited, and whose.
Whether the description was accurate.
Which competitors were named.
The date and model version.

MarketerHire frames the outcome as mentioned, recommended or cited, and measures visibility across mention rate, recommendation rate and sentiment.

It worked if someone else could repeat your run from the log. The typical mistake is logging only "yes" or "no" and losing the competitors and sources.

Report rates, not rankings

Turn the log into a rate, such as "mentioned in 7 of 15 runs." Skip rank positions. Rand Fishkin said any tool that gives a "ranking position in AI" is unreliable, while he considered visibility percentage across many prompts run multiple times a reasonable metric, per PPC Land. The SparkToro researchers likewise found that visibility percentage across many queries is more consistent than ranking position.

It worked if your report says "mentioned in X of Y runs" for each engine. The typical mistake is averaging engines together. Agility Writer notes that Perplexity cites sources aggressively, Gemini tracks Google's live index closely, and a brand can be the first pick in one and absent from another. Keep each engine on its own line. Citable also explains how to measure brand visibility in ChatGPT if you want to go deeper on metrics.

Read the prompts you lose

For each prompt where a competitor beats you, open the sources the answer cited. Agility Writer calls those pages your content gap list. Citable's article on why AI cites your competitors walks through this question by question.

It worked if you end with a short list of pages or topics to fix. The typical mistake is stopping at the score and never reading the cited pages.

Rerun on a schedule

Rerun the identical prompt set at a fixed interval and compare rates. Changing prompts between rounds breaks the comparison, so add new prompts as a separate group.

Can Google Search Console show when ChatGPT mentions you?

No. The Generative AI performance report is free, but eSEOspace says it reports impressions only: no clicks, CTR, position or list of questions. BrandJet adds that it doesn't show brand mentions in generated prose and doesn't cover ChatGPT or Claude.

Use it as a Google-only signal. Sonary lists it and Bing Webmaster Tools' AI Performance report as the two free starting points. To see visits that arrive from assistants, set up ChatGPT traffic tracking in GA4.

When should you move from a spreadsheet to a tool?

Manual sampling stops scaling at around 25 prompts or 2 competitors, whichever comes first, according to Agility Writer. Past that point, compare tools on the same criteria: cost, engine coverage, refresh cycle and main limitation.

Option	Cost	Coverage	Refresh	Main limitation
Manual log	Free	Any engine you test yourself	Whenever you run it	Stops scaling at around 25 prompts or 2 competitors
Search Console report	Free	Google only, impressions only	The sources consulted don't detail it	No brand mentions in generated text, no ChatGPT or Claude
OtterlyAI	From $29 a month on Lite, with 15 search prompts	Per Visiblie, no Claude, Grok, DeepSeek, Meta AI or Mistral	The sources consulted don't detail it	Google AI Mode and Gemini are paid add-ons
Ahrefs Brand Radar	The sources consulted don't detail it	Over 150M prompts across six AI indexes, per Ahrefs	Monthly for AI assistants	The sources consulted don't detail it
Citable	Starter $24/month per site, articles included	ChatGPT, Gemini, Perplexity and Google AI	Weekly	Claude and Grok are paid add-ons (Claude +$9/month on Starter; Grok +$9 on Starter, +$19 on Growth, +$35 on Business)

Which option fits depends on the situation. The manual log suits a small prompt set and a zero budget. Search Console suits a free Google-only impressions baseline. Ahrefs Brand Radar suits teams that want a very large prompt base and can live with monthly updates. Citable suits teams that want weekly runs plus articles and page fixes on top of tracking, but its Claude and Grok coverage costs extra, and teams that need those engines on the base plan should weigh that. Citable also looks into why competitors are picked instead of you, then writes sourced articles and makes page fixes, publishing them to your CMS. Whichever tool you choose, check that it reports rates across repeated runs.

Next steps

Write 20 buyer prompts across category, comparison, problem and recommendation intents, and drop any that only name your brand.
Run each prompt three times in fresh sessions on ChatGPT, Gemini and Perplexity, and log mentions, recommendations, cited sources and competitors.
Calculate a "mentioned in X of Y runs" rate per engine, then open the sources cited wherever a competitor won.
Check Search Console's Generative AI report for a Google-only impressions baseline.
If you later want the same measurement run weekly without the spreadsheet, Citable offers a free AI visibility report that shows whether ChatGPT mentions your site.

Frequently asked questions

How many prompts do I need to track brand mentions in AI?

Guides disagree. Omnia suggests 10 to 20 high-intent prompts as a baseline, while Ambeltek recommends 30 to 50 for most brands. Ambeltek also says fewer than twenty is too small a sample to show a trend.

How often should I check brand mentions in ChatGPT?

Rerun the same prompt set on a fixed schedule so rates stay comparable.

Can Google Search Console show when ChatGPT mentions my brand?

No. The report covers Google only, doesn't cover ChatGPT or Claude, and doesn't show brand mentions in generated text.

Is there a free tool to track brand mentions in AI answers?

There are two free starting points, according to Sonary: Google Search Console's Generative AI performance report and Bing Webmaster Tools' AI Performance report. Neither replaces a manual prompt log for mentions.

Why do AI answers change every time I ask the same question?

Outputs vary from run to run. In a SparkToro study, ChatGPT and Google's AI in Search returned the same brand list less than 1% of the time across repeated runs. That is why a rate across many runs beats a single result.

Sources

Keep reading

6 November 2026 · 9 min read

Step by step: how to monitor competitors in AI search

Build a fixed set of buyer prompts, run them on the AI engines your buyers use, and track share of voice and cited sources over time to monitor competitors.

5 November 2026 · 7 min read

FAQ pages that AI assistants quote: what the evidence shows

Write each FAQ heading as the question a buyer types and answer it in one or two standalone sentences. Schema alone showed no citation lift in tests.

4 November 2026 · 7 min read

Does Schema Markup Help in AI Search? What the Google and Ahrefs Data Show

Schema markup is not a proven lever for AI citations: Google says it isn't required and an Ahrefs test found no major uplift. Here is what to do.

Want this measured for you?

Get a free report on up to five buyer questions.

Get the free report