seo dispatcher

Google AI Overviews

How to Check Whether Your Website Appears in Google AI Overviews

Find your site's AI search appearances in Google Search Console, understand what the numbers mean, and choose a useful page improvement without guessing at AI rankings.

An illustrated AI answer card connects to source pages, with one source under a magnifying glass.

You search for a question your customers ask. Google puts an AI-written answer above the usual links. Other websites appear as sources. You wonder whether your site ever shows up there, or whether you need another tool to find out.

Start with Google Search Console's Generative AI performance report. It shows impressions for links to your site within Google's AI search features. Google announced the report in June 2026 and says its worldwide rollout finished on August 31. See Google's report announcement.

There is an important detail: the Search report covers AI Overviews and AI Mode. It is a view of appearances across those features, not a list of every answer that names your business. Here is how to use it without reading more into the figures than they show.

Open the report for the right website

Sign in to Search Console and select the property for your website. A property is the website address or domain you verified with Google.

Open the Generative AI performance report (Search). Google's help page links directly to the report if you have trouble finding it.

Choose a date range, then open Pages to see which addresses received impressions. Use Countries, Devices, or Dates when those views help answer your question. Keep the same filters when comparing periods. A change from mobile searches in one country to all searches would make the comparison less useful.

Pick one important page to investigate. You can export the report to keep a record of what it showed.

Understand what an AI search impression tells you

An impression records an appearance of a link. It does not mean the person opened your website, read the article, or became a customer.

The report announcement lists impressions and breakdowns by page, country, device, and date. It does not list a separate click count, CTR, or query table for this AI view. Do not calculate an “AI conversion rate” from appearances alone.

Keep three questions separate:

What you want to knowEvidence to use
Did links to our site appear in Google's AI search features?The Generative AI performance report.
Was our business named in a particular answer?Inspect that answer and record what you saw.
Did visitors complete a useful action on our site?Website analytics with the relevant action tracked.

The AI report's data is also included in the overall Search performance data. Adding those two reports together would count some appearances twice.

What if the report is missing?

Google says this usually means the site has not received enough impressions in its AI search features. A missing report is not proof that your website has an SEO fault.

Check Settings → Search generative AI too. Google provides an inclusion control there. Your property may follow a setting from a parent property, such as the main domain. If the site was deliberately excluded, understand that decision before changing it. Google's control guide explains the choices.

If your important page is missing from Google altogether, investigate that first. Our guide to a website not showing on Google explains how to read URL Inspection. Buying an AI tracker will not repair a page that cannot be reached.

Use a manual search to understand one answer

The report gives you a pattern. Looking at a real answer gives you context.

Choose a question a customer would actually ask. Write down the words you entered, the date, your country, and whether you checked an AI Overview or AI Mode. Note which websites were linked and whether your business was named.

Those are different observations. Google might link an article from your site without writing your company name in the answer. It might mention your company using information from another website.

Keep a screenshot or a short note for that check. Treat it as one observation, rather than a permanent ranking or a record of what every customer sees. This is also how to judge a tracker: ask which questions, places, and dates its results actually represent.

Follow one page from the report to a useful improvement

Imagine you sell stock-counting software for small shops. This is a made-up example. Your guide to doing a first stock count appears in the AI report, but you do not know whether those appearances led to visits.

Open the guide and try to follow it as a shop owner. It says to “organise your stock and ensure accuracy.” That sounds sensible, but what does the owner do next?

You could add a sample count sheet, explain what to record when the shelf count differs from the spreadsheet, and show how to check a second count. If your software helps with one of those steps, show the actual screen and explain what the user needs to enter.

You now have a reason for the update that stands on its own: the reader can finish a job they could not finish before. You have not proved that these changes will earn another AI appearance, and you do not need to invent that promise to make the article better.

Where SEO Dispatcher helps with that work

Add the public website to SEO Dispatcher, review its website profile, and audit the page you chose. Connect the matching Search Console property to examine the ordinary page and query data available in Search Insights. That can help you see whether the page serves the searches your business cares about.

Use Google's own report for the separate AI impressions check. SEO Dispatcher's standard Search Console views should not be read as a dedicated AI Overview report or a tracker of brand mentions across AI services.

Once you know what to improve, prepare a focused update in the editor. In our stock-counting example, ask for clearer steps and a worked count sheet, using only confirmed product features. Add your real screenshots and check the instructions yourself. The SEO Dispatcher workflow guide shows how to review a draft and publish supported Markdown or MDX content through GitHub.

Avoid adding things just to satisfy an AI checklist

Google's current AI search guidance says the foundations still matter: useful content, accessible pages, and clear explanations. A page must be indexed and eligible for a search snippet to be considered for its generative AI features. Meeting the requirements does not guarantee an appearance.

There is no special AI schema you must add. An llms.txt file is not a Google ranking requirement either. Our llms.txt guide explains when that optional document may serve another purpose.

Put the effort into the answer. Show how something works, use accurate examples, and remove claims you cannot support. If you use AI to help draft the update, review the content before publishing.

Check visits and useful actions separately

In GA4, the Traffic acquisition report can show visits from sources such as google / organic. It can also show actions you have configured as key events, such as a completed enquiry. Google's Traffic acquisition guide explains those views.

That source label alone does not identify which Google feature a visitor clicked. Keep the analytics figures beside your AI impressions, and resist treating them as a matched list of the same people. Search Console also does not report your appearances in ChatGPT or Perplexity; those need their own checks.

Record the page change and publication date. Return after enough data has collected, using comparable periods. Our SEO measurement guide helps you connect search observations with real business outcomes.

You should leave with two useful things: evidence of where your links appeared, and one clear way to improve the page for the next person who reads it.

See also

  • Google AI Overviews
  • AI search visibility
  • Google Search Console