How do you track brand mentions in AI search?

Short answer

Start with Google’s own data: Search Console reports how your pages perform in its AI features, which no third-party tool can see directly. Then run a fixed set of manual checks each month across the assistants your buyers use, and log the results. Treat paid “AI visibility” scores as rough estimates, not measurements.

A new category of tools has appeared promising to tell you how often AI assistants mention your brand. Some of them are useful. Most of them are measuring something narrower than they claim, and none of them has access to the data Google holds. Here is how I approach it for clients.

Start with Google’s own reporting

For Google’s AI features, the only first-hand data comes from Google. Its guidance on generative AI in search points site owners to the Generative AI performance report in Search Console for tracking how content performs in those features. Its documentation on AI features also confirms that sites appearing in AI Overviews and AI Mode are reported in the Performance report under the Web search type.

That is where I look first. It shows real impressions and clicks from real searches, rather than a sample of prompts someone chose to run.

Then run manual checks properly

For assistants outside Google, there is no equivalent report, so manual checking is still the most honest method. Done casually, it is misleading. Done consistently, it is useful.

Write a fixed list of prompts. Ten to twenty questions your buyers actually ask: “best criminal defense lawyer in [city]”, “how much does a roof replacement cost in [state]”, “who does Shopify SEO audits”. Keep the wording identical each month.
Run them in the assistants your buyers use. Logged out, or in a fresh session, so your own history does not shape the answer.
Record what you see. Whether you are mentioned, whether you are linked, who else is named, and which sources are cited.
Repeat monthly. AI answers vary from run to run. A single check means little. A trend over six months means something.

The cited sources are often the most useful part. They show you which pages and which sites the assistant leans on for your topic, and those are usually the places worth earning a mention or a link.

What the paid tools do

Most AI visibility tools run a set of prompts through one or more assistants on a schedule and count how often your brand appears. That is the manual method automated, at larger scale. It is useful for spotting trends across many queries.

The limits are worth knowing. Answers vary by user, location and session, so a tool’s sample is one view among many. The prompts are chosen by the tool or by you, not drawn from what real people asked. And none of them has access to actual usage data from the assistants. A “visibility score” is an estimate built on top of those assumptions.

I would not pay for one before looking properly at Search Console, and I would never treat its score as a business metric on its own.

Track the outcome, not just the mention

A mention in an AI answer matters if it leads to something. Watch for referral traffic from AI assistants in your analytics, where the assistant passes a referrer at all, and treat that number as a floor rather than a total. Ask new customers where they heard of you, and add AI assistants as an option on your enquiry form. Watch branded search impressions in Search Console, which tend to rise when more people encounter your name anywhere.

What not to do about it

Do not pay for mentions on sites built to be picked up by AI assistants. Google’s guidance says seeking inauthentic mentions isn’t as helpful as it might seem, and the pattern looks a lot like paid links did a decade ago.

Do not create llms.txt or similar files expecting Google to use them. Google says its search ignores them.

And do not rewrite your site in a special style for AI. Google says there is no separate optimisation needed to appear in its AI features. Pages that rank well, say something first-hand and are clear about who wrote them are what get cited.

A simple monthly routine

Check Search Console’s AI reporting and branded impressions. Run your fixed prompt list and log the results in a spreadsheet. Note any new cited sources worth approaching. Check analytics for AI referrals. Thirty minutes a month gives you a better picture than most dashboards.

What a good result looks like

For most of the businesses I work with, success is not being named in every answer. It is being named, accurately, in the handful of answers that matter for buying decisions in your area or niche, with the right description of what you do. If an assistant describes your business wrongly, look at the sources it cites. Usually one of them, often an old directory listing or an outdated page on your own site, is carrying the wrong information, and fixing that source fixes the answer over time.

If you want your brand mentioned where buyers and AI assistants both look, my link building service focuses on earning real coverage in the sources those answers actually cite.