How to Track AI Overviews and AI Mode in Google

How to track AI Overviews and AI Mode comes down to three separate measures: whether Google shows an AI answer for your keyword, whether that answer names or links your brand, and whether visits follow. Search Console now has a Generative AI performance report that shows impressions for AI Overviews and AI Mode, but it doesn't show whether an answer named your brand. The first two measures need a check by hand or a tracker that reads Google's results on a schedule.

So what can you see for free? What does a tracker add? And how many checks make a result you can trust?

Here's each method, what it can and can't tell you, and how to read the counts.

Key takeaways

• Track three things apart: an answer appears, it names or links you, and visits follow.
• AI Overviews and AI Mode are two Google surfaces, so count them separately.
• Search Console's Generative AI performance report shows impressions of your pages in both, but not whether an answer named your brand.
• A keyword with no AI answer is "not shown", which is a fact about Google and not a miss for you.

What should you track in AI Overviews and AI Mode?

Track whether an AI answer appears, whether it names or links your brand, and whether visits follow, separately for AI Overviews and for AI Mode. The two features behave differently, and one number for "Google AI" hides that.

Google says both features may use a technique called query fan-out, which issues several related searches across subtopics before building a response. It also says AI Mode and AI Overviews may use different models and techniques, so the answers and links they show will vary. Google adds that AI Overviews appear only when its systems judge them additive to classic Search, and that they often don't trigger.

That last point shapes everything below. If a keyword returns no Overview, nothing exists to name you, so the result describes Google's choice and says nothing about your brand.

Does Search Console show AI Overview traffic?

Search Console counts AI Overview and AI Mode clicks and impressions inside the Web search type of its Performance report, according to Google. Sites appearing in either feature are part of the overall search traffic there.

Google's Search Console help page gives the counting rules for impressions, position and clicks. A click on a link to an external page in an AI Overview or in AI Mode counts as a click, and standard impression rules apply to both. An AI Overview occupies a single position, and every link inside it takes that same position. In AI Mode, position follows the same method as a normal results page, and a follow-up question counts as a new query with its own impressions, position and clicks.

Those rules say how a click is counted in the overall report. Google also launched a dedicated Generative AI performance report on June 3, 2026, rolling it out to a subset of sites first. It shows how often URLs from your site appeared in AI Overviews and AI Mode, by page, country, device and date, so check whether it's available in your property.

The report counts impressions of your links. It doesn't say whether the answer's text named your brand, or which of your buyer questions triggered an answer. Search Console is the visit check, and the other two questions need another source. Google also notes that Search Console excludes data from experiments in Search Labs.

Google's guidance on AI features and your website adds that you can track conversions and time on site in a tool such as Google Analytics.

How do you check by hand?

Search each buyer question on Google from a US location and record whether an AI Overview or AI Mode answer appears, and whether it names or links your brand. It costs nothing and gives you a first read in an afternoon.

Follow these steps:

1. Write 20 buyer questions, phrased the way a customer would ask them. Include comparison and decision questions, not only definitions.
2. Search each one in a logged-out window from the same place each time.
3. Record whether an AI Overview appeared, whether it named your brand, whether its sources linked your site and which competitors it named.
4. Repeat each question in AI Mode as a single question, with no follow-ups.
5. Date the sheet, then repeat the whole set a week later.

Keep named and linked in separate columns. A link without your name in the text is a different result from a name with no link, and merging them hides which one moved.

Firemark's free checker does the reading for a short list. You enter a domain and up to five keywords, and it checks each one for the United States on desktop. It reports whether Google shows an AI Overview or an AI Mode answer and whether the answer links to your domain.

The tool is free and needs no sign-up. A keyword with no answer comes back as "not shown", not as a miss. Position is the order of the source links, because Google doesn't number its citations. A keyword checked in the last 7 days returns the earlier result.

The checker reads links, so it can't tell you whether the answer text names your brand. Five keywords is also a spot check. One look at one answer is one sample, and the next section explains why that isn't enough.

How to track AI Overviews on a schedule?

A scheduled tracker runs a fixed set of prompts on a calendar, reads each Google answer, and counts how often it names and links your brand. The schedule is what turns a one-off look into a trend.

AI answers vary from one ask to the next. SparkToro and Gumshoe had 600 volunteers run 12 prompts through ChatGPT, Claude and Google's AI Overviews (with AI Mode when no Overview appeared), 2,961 runs in all. Their study found less than a 1 in 100 chance that ChatGPT or Google's AI gives the same list of brands in any two responses, and less than 1 in 1,000 for the same list in the same order.

The study covers lists of brands and products, not citations, and it mixes three tools. Read it as a warning about single looks rather than a measure of Google alone. Its authors call visibility percentage across many runs a reasonable metric and suggest asking each prompt 60 to 100 times, a higher bar than one ask per prompt per run.

Any tracker worth using follows a few rules:

1. It asks a fixed set of prompts, at least 20 as a working floor here, so a share means something.
2. It asks from the same location and language every run.
3. It treats "no AI answer" as not shown, never as absent.
4. It counts named and linked separately.
5. It keeps AI Overviews and AI Mode apart, with no blended Google number.

Firemark applies rules 2 to 5. For AI Overviews, it runs each prompt as a Google search from the United States in English and reads the Overview on the results page, including one that loads after the page. Google decides whether a query gets an Overview, so the number it can read can be lower than the number of prompts.

For AI Mode, it sends each prompt as one question with no follow-ups and stores the answer with its sources. The two are counted on their own, with separate counts and histories.

Collection isn't always smooth. On its AI Overviews page, Firemark says that in its own early test runs this engine needed retries more often than its other four. Each prompt gets up to three attempts, and an engine that answers under 95% of prompts is left out of that run's totals, never counted as absence.

The size of the set decides what you can see. This arithmetic, from Firemark's method page, assumes a 40% share on one engine and is not data from a real brand.

At 25 prompts, a 5-point move sits well inside a 36-point range, so one run can't show it's real. The range covers sampling only, so it can't account for Google changing how it answers.

How do you read the results?

Read the share of readable answers that name you, and how that share moves over several comparable runs, instead of any single run. A count always needs its base, such as 12 of 40 answers.

Start with the shown rate. If few of your prompts return an AI Overview, that tells you how Google treats your category, and your share of the Overviews that do appear is the number to follow. Firemark's method page reports a movement only after three comparable runs and a change of 5 points or more. That's a house convention and not a statistical test, but it's a sensible floor to copy into your own sheet.

Then compare the two surfaces. A brand named in AI Overviews can be missing from AI Mode, and the reverse, because they can name different brands for the same question. Look at each on its own before you draw a conclusion about Google.

Finally, pair the counts with visits. Firemark records what answers say and doesn't measure clicks, visits or sales, so use Search Console and analytics for that. If you need traffic attribution or keyword rank history, those tools fit better than a visibility tracker.

If you'd rather pick a tool than build a sheet, two comparisons set the options side by side: AI Overview trackers and AI Mode trackers. Firemark measures and doesn't change what Google shows. Start with 20 buyer questions and a sheet, or open the demo report to see counts on sample data.