Clients now ask a question that rank reports cannot answer: "When people search our category, does Google's AI Overview mention us?" A page can hold the first organic position and still lose most of its clicks to the overview above it. Ahrefs estimated, using December 2025 data, that an AI Overview reduces the click-through rate of the first result by about 58%.
An AI Overview audit answers that client question and turns it into a work plan. This checklist has ten checks in four groups. Each one says what to look at, how to check it, and when you can call it done.
Before You Start: Get a Baseline
Record where the site stands before you change anything; otherwise you can't tell later whether a fix did anything.
1. List the searches that matter
Write down 15 to 30 searches a buyer would type before they know the brand. Use three kinds:
- Category searches: "best accounting software for freelancers", "top digital agencies in Jakarta".
- Comparisons: "Xero vs QuickBooks", "in-house SEO vs agency".
- How-to questions close to the product: "how to file quarterly taxes as a freelancer".
Leave the brand name out. A search for the brand itself almost always cites the brand's own site and tells you nothing about how new customers find it.
You're done when the client agrees the list describes how their buyers search, grouped by product line or market.
If the client sells in more than one country or language, build a separate list for each market and write the searches the way local buyers phrase them. AI Overviews are generated per market, so an English search run from the US says little about the same question asked in Bahasa Indonesia from Jakarta. Run each market's searches from that country.
2. Record what the overviews show today
For each search, note whether an AI Overview appears, which brands it names, and which pages it links as sources. Overviews change from one day to the next and differ by country. Google also shows them only when its systems decide they add something to the regular results, so many searches have none.
You can do it by hand in a private browser window, set to the country you sell in. For a first pass across a whole category, a tool saves time: AskWatch's free checker shows which pages Google AI Overviews cite for buyer searches in your category, along with the brands they name and where your brand sits in the answer.
The output is a dated table with one row per search and four columns: overview shown, brand named, site cited, competitors named.
Can Google Use Your Pages?
Google's guidance on AI features is short on this point: to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to appear in Google Search with a snippet. There are no extra technical requirements. Checks 3 to 5 confirm that the basics are in place.
3. Crawl access for Googlebot
Open the site's robots.txt and check the rules for Googlebot and for all user agents. A blanket Disallow on a folder that holds guides, documentation or comparison pages keeps those pages out of everything Google builds, including the overviews.
Then check the layer in front of the site. Content delivery networks and security plugins sometimes challenge or block crawlers that robots.txt allows. Google lists "crawling is allowed in robots.txt, and by any CDN or hosting infrastructure" among the basics for AI features. The URL Inspection tool in Search Console shows whether Googlebot can fetch a page and what HTML it received.
Every page you want cited should return a normal page to Googlebot, with no challenge screen and no blocked resources that hide the main text.
4. Indexing and snippet controls
A page that is crawlable but not indexed cannot be cited. In Search Console, check the Page indexing report for the pages on your baseline list. Common reasons for exclusion are a noindex tag left over from staging, a canonical tag pointing to another URL, and thin duplicates of the same page.
Next, look for snippet controls. The nosnippet and max-snippet robots directives, and the data-nosnippet attribute on page sections, limit what Google can show from a page. They apply to AI features as well as to regular snippets. Teams sometimes add them to protect pricing or gated content and forget they are there.
To find stray data-nosnippet attributes, search the page source for the string, or set up a custom search in your site crawler so it lists every page that contains it.
At the end of this check, every page on the baseline list is indexed, has its own canonical, and carries no snippet limits that nobody chose on purpose.
5. Main content in text form
Google asks that important content be available in textual form. Check the pages for answers that live only in images (a pricing table exported as a picture, an infographic with the key numbers), inside tabs that load only on click, or behind scripts that fail for crawlers. Use the rendered HTML from URL Inspection to confirm the text Google receives.
Pass this check when the facts a buyer asks about (what the product does, who it is for, prices, comparisons) appear as text in the rendered page.
Is the Answer Easy to Lift?
Once Google can use a page, the question is whether the page answers the search well enough to be quoted. Google describes a "query fan-out" technique: the system runs several related searches across subtopics and picks supporting pages for each. A page can be cited for one of those sub-questions even when it does not rank first for the original search.
6. Put the answer at the start of each section
Look at how each priority page opens its sections. In the overviews we have reviewed, cited pages usually answer the question in the first lines: a one-sentence definition, a short list of options, a direct comparison with the deciding factor named. Pages that spend three paragraphs on background before the answer give the overview less to work with.
How to check: read the first two sentences under each H2. If they do not answer the heading on their own, rewrite them so they do, and move the background below.
When you finish, every section reads as a short answer to its own heading.
An example. A section headed "How much does bookkeeping software cost?" that opens with the history of cloud accounting is hard to quote. The same section opening with one sentence that gives the price range and names what drives it, such as the number of users or bank connections, gives the overview a sentence it can quote.
7. Cover the follow-up questions
Buyers rarely stop at one question. Someone comparing accounting tools will also ask about price per user, integrations with their bank, and what happens to their data if they switch. A page that covers them has more chances to match one of the searches Google runs in the background.
How to check: for each search on the baseline list, write the three to five questions a buyer would ask next. Mark which ones the page answers. Search Console's Queries report for the page often shows the follow-ups people already type.
The target: each priority page answers the main question and its common follow-ups, each under its own heading.
8. Keep facts current and dated
If a competitor's page lists this year's prices and yours lists last year's, you have given Google an easy reason to cite theirs. Discontinued features and statistics from three years ago work the same way.
How to check: list every number, price and feature claim on the priority pages and note when it was last confirmed. Add a visible "updated" date to pages that change often, and update it only when the content changes.
Done means every price and feature claim matches the current offer, and every statistic carries a source and a date from the last 12 months.
What Do Others Say About You?
For buying questions, overviews often cite review sites, comparison articles and forums rather than the brands' own pages. Part of the audit therefore looks beyond the client's site.
9. Third-party pages that keep getting cited
Go back to the baseline table and list every source page the overviews linked. Group them by type: the client's own pages, competitors' pages, review platforms, media and comparison articles, forums and community threads, and video. The baseline table will usually show the same few third-party pages across many searches.
Then check what each of those pages says about the client. Is the brand listed at all? Are the price, features and positioning correct? Is the review profile complete and recent? A common case: a review profile still lists a plan the client dropped a year ago, or a price from before the last increase, and the overview repeats it.
End with a short list of the third-party pages that matter in this category, the client present and described accurately on each, and an owner for each fix.
The fixes differ by type. A review platform usually needs a complete, current profile and a steady flow of reviews from real customers. A comparison article needs a pitch to its editor with accurate product facts. A forum thread needs an honest answer from someone at the company, posted under their own name, where the question is still open.
10. Consistent brand facts everywhere
If the website, the Google Business Profile, the Merchant Center feed and review profiles describe the client differently, the answer can mix old and new facts.
How to check: compare the one-line description, product names, prices, locations and opening hours across the site and the main profiles. Google's own guidance for AI features asks site owners to keep Merchant Center and Business Profile information up to date, and to make sure structured data matches the visible text on the page.
The check passes when the site, the profiles and the structured data give one set of facts in the same words.
Running the Audit as a Routine
The results from check 2 shift from week to week, so the baseline needs repeating. A practical cadence:
- Every two to four weeks: rerun the baseline searches and compare who is named and cited with the previous run.
- After each fix: note the date and the page, then check the related searches in the following runs. Judge a change by the pattern over several runs.
- Every quarter: review the search list itself. Add new products, new competitors and new questions the client's sales team hears.
Search Console helps with part of this. Its generative AI report, available to all properties since 31 August 2026, shows how often the site's URLs appeared in AI features, broken down by page, country, device and date. It does not show the searches behind those impressions, the brands named in the answers, or which competitors were cited, so the baseline table is still needed. Agencies that run this for several clients often track the same searches daily with a dedicated tool such as AskWatch, which records the answers from Google AI Overviews and other AI engines over time.
For reporting, add "cited in the AI Overview" as its own line next to rankings and clicks. A page that ranks first and is cited in the overview may still lose clicks, because fewer people click any result. Without that line, a click drop on a page that ranks first looks like an SEO failure.
A Short Version for the Client
When you present the audit, the client needs three things on one page:
- Where they stand: how many of their buyer searches show an overview, how many name them, and how many cite their site.
- Why: the main blockers found in checks 3 to 10, in order of how many searches they affect.
- What happens next: the fixes, who owns each one, and the date of the next baseline run.
Attach checks 1 to 10 as an appendix the client's team can work from, and send the next baseline run on the date you agreed.
About the author: Max is Head of Growth at AskWatch, which tracks how AI search engines such as Google AI Overviews, ChatGPT and Perplexity mention and cite brands.
Sources & References:
- Update: AI Overviews Reduce Clicks by 58%. Ahrefs, February 2026. ahrefs.com
- AI Features and Your Website. Google Search Central. developers.google.com