AI search
How to track your brand in Google AI Overviews (and which tools do it)
How to check whether Google's AI Overviews and AI Mode name and cite your brand, by hand and with tools, and how to pick the right one.
Track Google AI Overviews by logging, per search and market, whether an Overview appears, whether it names you, and which pages it cites.
Here is the direct answer. Start free: Google Search Console now has a generative AI report that shows which of your pages appeared in AI Overviews and AI Mode. Then check 10 to 20 searches by hand, logged the same way each time. When you need competitors, trends and more than one market, use a tool. Free checkers from Ahrefs and Semrush give you a one-off read. SEO suites add AI tracking to the rankings you already watch. Dedicated AI visibility platforms such as Otterly.AI, Peec AI, Profound and 3RD track AI Overviews and AI Mode next to ChatGPT and the other assistants. Only a few go on to say what to fix.
Your SEO dashboard says you rank third for your most important search. That tells you nothing about the AI Overview sitting above that result. The Overview may name two competitors and cite a review site that has never heard of you. Nothing in your rankings report records that moment, and a buyer who reads the Overview and stops there never clicks through to your site.
This guide covers the job in order: what Google's AI answers show, why they behave differently from ChatGPT, how to track them by hand, where the manual method breaks, and which tools do it. The worked example comes from a 3RD demo workspace, not a customer account.
The short version
|
Step |
What you do |
Rule of thumb |
|---|---|---|
|
1. Pick the searches |
Choose the keywords and questions your buyers type into Google |
10–20 to start, split into short keywords and full questions |
|
2. Read Search Console |
Open the generative AI report for your pages |
Free, but it shows your linked pages only, not mentions or competitors |
|
3. Check by hand |
Run each search in a clean browser, per market |
Log every check the same way, on at least three different days |
|
4. Know the limits |
Personalization, location and constant change |
A manual log shows big patterns, not small movements |
|
5. Pick a tool |
Match the tool to the job: checker, suite add-on or platform |
Ask how it collects answers and whether it tracks AI Mode too |
|
6. Act on the gaps |
Fix your pages and win the third-party pages Google cites |
Treat Google's AI surfaces as one job |
For the general method across all assistants (prompt types, tiers, sample sizes), read how to measure your brand's visibility in ChatGPT, Perplexity and Google AI Overviews. This page stays on Google.
What AI Overviews and AI Mode actually show
Google runs two AI answer surfaces in Search, and you should track both.
AI Overviews are the summaries at the top of a normal results page. Google shows one only when its systems judge that it adds something to the classic results. Many searches don't trigger one at all. When an Overview appears, it names brands in the text and links to supporting pages, often behind a "show more" or a link icon.
AI Mode is a separate, conversational search mode. Google aims it at questions that need "further exploration, reasoning, or complex comparisons". Buyers can ask follow-up questions, and each follow-up counts as a new search.
Google says both may use "query fan-out": the system runs several related searches across subtopics and sources, then writes one answer from what it finds. So the pages cited in an answer don't always rank for the exact words the buyer typed.
For each search, there are five things to record:
|
What to record |
Why it matters |
|---|---|
|
Overview shown? |
No Overview means no AI answer to win on that search. Track the rate, because it changes. |
|
Brand named? |
A mention in the text is what the buyer reads. It can happen with no link to you. |
|
Your pages cited? |
A citation can send a click, and it shows which page Google trusts. It can happen without a mention. |
|
Position and competitors |
First pick and fifth in a list are different results. Note who is named instead of you. |
|
Accuracy |
Old prices, retired products or the wrong company with a similar name. Being named wrong is worse than not being named. |
Mentions and citations are separate signals. A page can be cited while the answer names a competitor, and your brand can be named while the links go to a review site. Track both, or you'll misread the result.
Why Google's AI answers differ from ChatGPT
If you already check ChatGPT, don't assume the Google numbers will match. They often don't. Three differences drive most of the gap.
- They draw on Google's index. AI Overviews and AI Mode build answers from pages Google has indexed and can show with a snippet. Your classic SEO work (crawlable pages, internal links, structured data that matches the visible text) still decides whether a page is eligible.
- They don't always answer. ChatGPT answers every question. An AI Overview appears only on some searches, and whether it appears can change from week to week. So "how often are we named" needs a second number next to it: how often an Overview appeared at all.
- They lean more on your own pages. Across 100 brands and 5.9 million citations in 3RD's data, Google's AI surfaces cited the brand's own site more often than the chat assistants did. Owned pages took a median 12.0% of citations in AI Mode (20 brands) and 9.7% in AI Overviews (50 brands), against 5.8% in ChatGPT (97 brands) and 3.2% in Claude (32 brands). That still means about nine in ten citations go to other people's pages. But on Google, your own content carries more weight than it does in a chat assistant.
There's good news too: Google's surfaces move together. Across 207 brands tracked in three or more assistants, the typical gap between AI Mode and AI Overviews was 2.5 points, and between Gemini and AI Overviews 3.7 points. Between Copilot and Gemini, it was 15 points. So treat AI Overviews and AI Mode as one job, not two. We cover the full assistant-by-assistant picture in your weakest AI assistant shows your brand half as often as your strongest.
Step 1: Pick the searches your buyers type
AI Overviews and AI Mode get different kinds of input, so your list needs two kinds of searches.
- Short keywords for AI Overviews: "payroll software", "best running socks", "crm for agencies". This is how most people still type into Google, and it's where Overviews appear above the results you already rank for.
- Full questions for AI Mode: "what's the best payroll software for a 200-person company in Denmark?" People talk to AI Mode the way they talk to ChatGPT.
Where to find them:
- Search Console queries. Take the queries you already get impressions for. Keep the short ones as keywords, and rewrite the longer ones as questions.
- Your sales calls. The questions prospects ask before buying are the questions they now type into AI Mode.
- "Best X for Y" searches. These decide who makes the shortlist, and they trigger comparisons in AI answers.
Keep branded and unbranded searches apart. A search that contains your brand name almost always gets an answer about you. It tells you whether Google describes you correctly, not whether new buyers find you. Report the two as separate numbers.
Start with 10 to 20 searches. Choosing the right ones matters more than adding volume.
Step 2: Read Search Console's generative AI report first
Before you build a spreadsheet, open what Google already gives you. In June 2026, Google added generative AI performance reports to Search Console. As of August 31, 2026, Google says they are available to all websites.
What the report shows: impressions, meaning how often URLs from your site appeared in generative AI features in Search (AI Overviews and AI Mode) and in Discover. You can break them down by page, country, device (Search only) and date.
What it doesn't show, based on Google's announcement:
- Which searches triggered the impression. The announcement lists pages, countries, devices and dates, not queries.
- Mentions without a link. If an Overview names your brand and links to a review site, your site gets no impression.
- Competitors. You see your own site only.
- A split between AI Overviews and AI Mode. The announcement groups them as "generative AI features".
The main Performance report also still counts these features inside the "Web" search type. A click on a link in an AI Overview counts as a click. An AI Overview takes up one position in the results, and every link in it shares that position. So in the main report, a page cited inside an Overview can look like it ranked first.
Rule of thumb: Search Console tells you which of your pages Google's AI features use. It can't tell you whether Google recommends you over a competitor. Use it to find the pages worth protecting, then check the answers themselves.
Step 3: Check AI Overviews and AI Mode by hand
You can get a first read in an afternoon. The method only works if you run it the same way every time.
The routine:
- Open a clean browser. Use a private window, signed out of Google. Search history and signed-in personalization can change what you see.
- Set the market. Search in the market's language. To check another country, test from that country with a VPN, or use a colleague who is there. Write down which method you used.
- Run the short keywords on the normal results page. Record whether an AI Overview appears. If it does, expand it fully before you read it, and open the source links.
- Run the questions in AI Mode. Open AI Mode from the results page and ask the full question. Ask only the first question. Follow-ups are new searches and make your log hard to compare.
- Repeat on at least three different days. Google's answers change between checks, so one day is an anecdote.
- Log every check the same way.
Log these columns:
|
Column |
Example |
|---|---|
|
Date, surface, market, device |
2026-11-16, AI Overview, US–English, desktop |
|
Search |
"best running socks" |
|
Overview shown? |
Yes / No |
|
Brand named? |
Yes / No |
|
Position in the answer |
3 of 5 |
|
Competitors named |
Brand A, Brand B |
|
Your pages cited |
/guides/running-socks |
|
Third-party pages cited |
A review site's "best running socks 2026" list |
|
Accurate? |
No, it quotes last year's price |
Take a screenshot of every answer that names a competitor and not you. Those are the answers you'll want to reread when you plan the fix.
The limits of tracking by hand
A manual log is a good start and a weak measurement. Know where it breaks before you report it.
- Personalization. Signing out reduces it but doesn't remove it. Your location, language and device still shape the answer.
- Location. AI Overviews differ by country and language. A VPN gets you close, but you can't be sure you're seeing what a buyer in Ohio or Copenhagen sees.
- Volatility. Overviews appear, disappear and change wording between checks. Across 111 brands in 3RD's data, the median brand's daily visibility moved 12.5 points between its best and worst day in one month, with no change to its prompts. A manual check can't tell a real change from that kind of swing.
- Triggering. On some days an Overview won't appear for a search where it appeared last week. If you don't log "Overview shown?", a missing Overview looks like a lost mention.
- Hidden sources. Overviews collapse, and links sit behind icons and "show more". If you don't expand everything, you'll under-count citations.
- Time. Twenty searches on two surfaces over three days is 120 checks a market. Most teams do it once, and the routine slips by the second month.
Once you need competitors, more than one market and a trend you can defend, it's time for a tool.
Which tools track brand mentions in AI Overviews and AI Mode
The tools fall into four groups. They differ less in whether they "see" AI Overviews than in four practical questions:
- Which surfaces? AI Overviews only, AI Mode only, or both, plus the other assistants?
- Mentions or links? Does it count brand names in the text, or only links to your domain?
- Which markets? Can you set the country and language, and how many?
- What happens after the dashboard? Does it explain why a competitor gets picked and help you do the work, or does your team start from scratch?
Also ask how each tool collects answers. Some capture the live Google results page, some run fixed prompts through a provider's API, and some model visibility from a large keyword database. All three are valid. They answer slightly different questions, so ask which method a tool uses for Google.
The decision table
All vendor facts are as of November 2026, from each vendor's own pages. "Not stated" means we couldn't find it on the pages we checked.
|
Tool |
Google surfaces covered |
What it counts |
Markets |
Published price |
What happens after |
|---|---|---|---|---|---|
|
Google Search Console (generative AI report) |
AI Overviews and AI Mode, grouped together |
Impressions of your own URLs |
By country |
Free |
Nothing. Your team reads and decides |
|
Ahrefs free AI Overviews tracker and AI Mode tracker |
One tool per surface |
Mentions, top cited domains and pages |
Not stated; "strongest in English" |
Free, limited preview |
Links to Brand Radar |
|
Ahrefs Brand Radar |
AI Overviews, AI Mode, plus ChatGPT, Perplexity, Gemini, Copilot and Claude |
Mentions and citations, from a large prompt database plus custom prompts |
Not stated on the pages we checked |
Paid add-on; AI Visibility Index from $199/month, custom prompts from $50/month |
Research data; your team acts |
|
Semrush free AI Overviews checker |
AI Overviews |
Mentions, cited pages, a relative visibility score |
Not stated |
Free |
Links to paid tools |
|
Semrush AI Visibility Toolkit |
Prompt tracking names Google AI Mode and ChatGPT |
Visibility score, share of voice, sentiment |
Not stated |
$99/month; no free trial for the toolkit |
Recommendations and a site audit |
|
SE Ranking AI visibility tracker |
AI Overviews, AI Mode, ChatGPT, Gemini |
Mentions with and without links |
7 supported markets, 5 languages |
Not on the tracker page; 14-day free trial |
Competitor and source research |
|
Keyword.com AI Mode tracker |
AI Mode (a separate tracker covers AI Overviews) |
Mentions, sentiment, influential citations |
Not stated |
Not shown on the page; 14-day free trial |
Full scraped answer per check |
|
Otterly.AI |
AI Overviews, AI Mode, plus ChatGPT, Perplexity, Copilot, Gemini and Claude |
Mentions, citations |
"Unlimited" countries on every plan |
From $29/month (16 prompts), credit-based |
Recommendations and audits |
|
Peec AI |
Choose 3 models from a list that includes AI Overview, AI Mode and Gemini |
Visibility, position, sources |
1 country on Starter, 3 on Pro |
From $95/month (50 prompts) |
Your team acts on the data |
|
Profound |
Enterprise only: ChatGPT, Gemini, AI Overviews and AI Mode |
Visibility, citations |
Custom on Enterprise |
Enterprise only, custom price; no free trial listed |
Agents draft content for approval |
|
3RD |
AI Overviews and AI Mode among the models you choose, next to ChatGPT, Gemini, Copilot, Perplexity and Claude |
Mentions, position, citations, sentiment, Recommended rate |
Any country–language pair you add |
From €99/month (up to 2 models); 7-day free trial, no card |
Diagnosis, a ranked plan, and drafts for you to check and publish |
What each option is good at
Google Search Console. The only source with Google's own impression data, and it's free. Use it to find which of your pages Google's AI features already use. It can't show mentions, queries or competitors, so it can't tell you whether you're winning.
Ahrefs. The free AI Overviews and AI Mode trackers are a fast first look, with no signup. Brand Radar is strong if your team already lives in Ahrefs: it runs a very large set of search-backed prompts, which suits research on a whole category. Ahrefs itself describes some of its share-of-voice numbers as modeled visibility, meaning potential exposure rather than measured reach. That's a fair description for a research tool.
Semrush. The free AI Overviews checker shows mentions, cited pages and the searches where competitors appear without you. Semrush's own page warns that the score is relative and "can change from week to week". The paid AI Visibility Toolkit names Google AI Mode and ChatGPT for prompt tracking. It's the natural add-on if your SEO reporting already runs in Semrush. If you're weighing a suite add-on against a separate platform, read do you need a separate tool for AI search, or does your SEO tool cover it?
SE Ranking. Its tracker counts mentions both with and without links to your site, which is the right split for AI Overviews. Coverage is four assistants, two of them from Google, across seven supported markets.
Keyword.com. A focused AI Mode tracker, with the full captured answer kept for every check. Good if AI Mode is your single question.
Otterly.AI, Peec AI and Profound. Dedicated AI visibility tools that put Google's surfaces next to ChatGPT and the rest. Otterly.AI covers seven engines and prices them in credits. Peec AI lets you choose three models on its self-serve plans, from a list that includes AI Overview and AI Mode. Profound now sells Enterprise only, which covers AI Overviews and AI Mode next to ChatGPT and Gemini. Detailed comparisons are in our Otterly alternatives, Peec AI alternatives and Profound alternatives guides.
3RD. 3RD tracks AI Overviews and AI Mode as two of the models you choose, in every country and language you add, and scores whether the answer recommends you over the alternatives, not only whether it names you. Then it explains why a competitor gets picked and turns that into a ranked plan. 3RD Agent drafts the work, and your team checks and publishes it. More on this in the worked example below.
Which one fits your situation
|
Your situation |
Start with |
Why |
|---|---|---|
|
You want to know if Google's AI uses your pages at all |
Search Console's generative AI report |
Free, and it's Google's own data |
|
You need a one-off read for a meeting this week |
A free checker (Ahrefs or Semrush) plus a manual log of 10–20 searches |
Fast, no contract |
|
Your SEO team lives in one suite and Google is the only AI surface you care about |
That suite's AI add-on |
One login, next to the rankings you already report |
|
AI Mode is the one surface leadership asks about |
A focused AI Mode tracker |
Narrow job, narrow tool |
|
Buyers also use ChatGPT and other assistants, in more than one market |
A dedicated AI visibility platform |
Google and the chat assistants side by side, per market |
|
You have no AI visibility specialist and need the fixes done |
A platform that diagnoses and drafts the work, such as 3RD |
The tool does the analysis and the first draft; your team checks and publishes |
For the full buying checklist across all assistants, see how to choose an AI visibility platform. For prices and models across more tools, see the best AI visibility tools in 2026.
Worked example: Google's AI surfaces in a demo workspace
Here is the job in 3RD, in a demo workspace: a sportswear brand tracked against six competitors in Denmark, Sweden, the UK and the US, across seven AI models and 207 prompts. Brand names are blurred, and the numbers are not customer results.
1. Split the score by model. The blended score hides the Google problem. Over the past week, ChatGPT named the brand in about 60% of answers. Google AI Mode named it in about 20%, and AI Overviews in about 21%. The two Google surfaces sit within a point of each other, and far below everything else.
Visibility by model for the past week. AI Mode and AI Overviews move together, about 40 points below ChatGPT. Demo workspace; brand names blurred.
2. Ask what to fix first. The team asks 3RD Agent: "What should we focus on first to get chosen more often when buyers ask AI which running shoes to buy?" The brand is mentioned in 50% of answers but recommended in 1.8%, second behind a leader at 5.4%. Claude and DeepSeek recommend it most (3.8% and 3.7%). Gemini and AI Mode almost never do (0.3%). Its read is "a content and positioning problem, not a reach problem". The second of its three priorities is to close the gap on Gemini and AI Mode specifically.
3RD Agent's answer: Google's surfaces are where the brand is least often recommended, so they get their own priority. Demo workspace.
3. Find the pages Google's answers lean on. In Citations, the team looks at the sources AI cites for its prompts, then filters by model to see what Google's surfaces use. Across all models, the most-cited sources include a running-socks roundup and a news site's workout-socks list that don't mention the brand at all. Those pages are outreach targets. And because Google's surfaces cite owned pages more than the chat assistants do, the brand's own buying guides are worth fixing first for AI Overviews.
The pages AI cites for the brand's prompts, across all models, and whether each one mentions the brand. Several lists name other brands but not this one. Demo workspace.
From there, the brand sets a goal for its Google gap, gets a Playbook of ranked tasks, and has 3RD Agent draft the page rewrites and outreach. Someone on the team checks each draft and publishes it. 3RD then reads the targeted prompts against the brand's other prompts, so the team can see whether the change moved more than the normal drift.
What to do once you can see the gap
Tracking is the easy half. These moves address what usually causes a Google gap:
- Make your best pages easy to quote. Google's AI surfaces cite owned pages more than chat assistants do. Put the answer first, use clear headings and tables, and state facts plainly: prices, specs, who it's for.
- Fix what's wrong. If an Overview quotes an old price or a retired product, find the page it cites and update it, whether it's yours or a third party's.
- Win the third-party pages Google cites. About nine in ten citations go to other people's pages. The roundups and review sites that name competitors without you are a short, concrete outreach list.
- Check eligibility. A page has to be indexed and eligible for a snippet to appear as a supporting link. If a page is blocked, unindexed or thin on text, Google's AI features can't use it.
- Work on AI Overviews and AI Mode together. They draw on the same index and move together. One fix usually moves both.
Common mistakes
- Counting links only. Search Console and many SEO tools see your URLs. They miss the answers that name you and link to someone else, and the answers that name a competitor while citing your page.
- Treating "no Overview" as "not mentioned". Log whether an Overview appeared. Otherwise a quiet week for Google looks like a loss for you.
- Reading one day. Overviews change between checks. Compare two- to four-week windows, not two screenshots.
- Checking only from your own office. Your country, language and account shape the answer. Set the market deliberately and record how.
- Tracking AI Overviews and ignoring AI Mode. Buyers who want to compare options are the ones Google points to AI Mode. Track both.
- Reporting Google as "our AI visibility". Google's surfaces can sit far below or far above ChatGPT. In the demo workspace, the gap was about 40 points. Report each assistant on its own line.
- Stopping at the dashboard. A Google gap has causes: thin pages, missing facts and third-party lists that skip you. If nobody owns the fix, the number won't move.
What 3RD doesn't do
- It doesn't replace Search Console. Google's own impression and click data for your pages lives there. Connect Search Console and GA4 to 3RD to see the traffic side next to visibility.
- It doesn't promise it sees exactly what one person sees. 3RD collects answers through each provider's API, with the market set as context for every question. Treat its numbers as a consistent sample across markets and days, and spot-check a few searches by hand.
- It doesn't make Google show an Overview, or change a third party's page for you. It shows which pages matter and drafts the pitch.
- It doesn't publish for you. 3RD Agent drafts, and your team checks and publishes.
- It doesn't promise speed. Expect days to weeks between publishing a change and a measurable shift.
About the data
- Owned citations by assistant: from The state of AI visibility in 2026. 100 customer brands and 5,926,953 citations, September 19 to October 18, 2026. Median owned share per assistant, among brands with 500 or more citations on that assistant (AI Mode n = 20, AI Overviews n = 50, ChatGPT n = 97, Claude n = 32). Model coverage differs by brand, so treat the split as directional.
- Gaps between assistants: from your weakest AI assistant shows your brand half as often as your strongest. 207 customer brands with at least 200 answers in each of three or more assistants, October 4 to November 2, 2026. The typical gap is the median absolute difference across brands tracked in both assistants.
- Daily swings: 111 customer brands and 732,000 AI answers, September 19 to October 18, 2026.
- Caveats: the data is observational, anonymized and aggregated. No brand is named. The sample leans toward Nordic brands, and customers choose which assistants they track. The screenshots and worked example come from a 3RD demo workspace, not a customer.
- Vendor facts: from each vendor's own pages and Google's own documentation, checked November 12, 2026. Prices and plans change, so check before you buy.
FAQ
Can I track my brand in Google AI Overviews for free? Yes. Search Console's generative AI report shows which of your pages appeared in AI Overviews and AI Mode, by page, country, device and date. For mentions and competitors, log 10 to 20 searches by hand in a clean browser, on at least three days. Free checkers from Ahrefs and Semrush add a one-off read.
Does Google Search Console show AI Overview mentions? No. It shows impressions of your own URLs in generative AI features, grouped across AI Overviews and AI Mode. It doesn't show which searches triggered them, answers that name you without linking to you, or competitors. In the main Performance report, AI Overview clicks and impressions sit inside the "Web" search type.
What's the difference between tracking AI Overviews and AI Mode? AI Overviews appear above normal results, and only on some searches, so track short keywords and record whether an Overview appeared at all. AI Mode is a conversational search mode, so track full questions. In 3RD's data the two sit about 2.5 points apart for the typical brand, so work on them together.
Why does my brand show up in ChatGPT but not in AI Overviews? They read different sources. Google's surfaces build on Google's index and lean more on brands' own pages than ChatGPT does. Compare the pages each one cites for the same question. The difference usually shows what to fix.
How often should I check AI Overviews? Continuously, if you can, and read the results over two- to four-week windows. Overviews change between checks, and a single day can mislead. A monthly manual check catches only large shifts.
Start this week
Pick your ten most important buyer searches. Open Search Console's generative AI report, then run those searches in a clean browser on three different days, and log the columns above. That gives you a baseline and a list of the pages Google trusts instead of yours.
If you want AI Overviews and AI Mode tracked next to ChatGPT and the other assistants, in every market you sell in, with the fixes drafted for you, start a 7-day free trial. No card needed. Or book a demo and we'll walk through your Google gap with you.


