What Is an AI Overview?

AI Overviews are Google's AI-written summaries that can appear at the top of search results and give a quick overview of a topic with links to learn more. For a health practice, they sit between the patient's question and your website, so the summary may be read before any result is clicked. Google began expanding them to all U.S. users in May 2024.

How AI Overviews work

When a patient types a question into Google, the usual page of results can now begin with a block of text written by AI. Google says AI Overviews run on a custom version of its Gemini model built for Search, which combines multi-step reasoning with Google's Search systems. Beside or under the text sit links to pages the system used or that support what it wrote.

Four details matter for a practice.

• Google decides when one appears. An overview shows up only when Google's systems decide it adds to classic Search. Many searches still show only the familiar results.
• The links come from normal Search. Google says a page must be indexed and eligible to be shown in Google Search with a snippet to appear as a supporting link, and it lists no additional requirements.
• It can run several searches behind one question. Google calls this technique query fan-out. The system issues multiple related searches across subtopics and data sources, then builds one response. A page can therefore be linked for a subtopic the patient did not type.
• Your usual reports include it. Google says traffic from sites in these features is included in overall search traffic in Search Console and appears under the Web search type.

On the question of special tags, Google is clear. It says you do not need new machine readable files, AI text files or markup, and there is no special schema.org structured data to add. You control access the way you do for any search, with robots.txt for crawling and snippet controls such as nosnippet, data-nosnippet, max-snippet or noindex if you want to limit what shows.

It is worth putting the scale in context. Google reported that AI Overviews drove a more than 10 percent increase in usage of Google for the queries that show them in markets such as the U.S. and India, based on its internal experiment data from September 2024 to April 2025. That is Google's own figure and the company has a stake in the result, so treat it as direction and not as independent evidence. It does suggest these summaries are being used, not ignored.

AI Overviews are related to older features. The featured snippet is a boxed answer taken from one page. An overview is written fresh and links several pages. Google's separate AI Mode is a full conversational tab with follow-up questions, built on the same query fan-out idea.

For a practice, the consequence is behavioral. A patient may read the summary, see a few names, and search one of them later. That is the pattern behind zero-click search, where value is delivered on the results page. Being linked, and being named in the text, both count.

One practical consequence is that your monitoring has to widen. A rank tracker that reports your position for one keyword cannot tell you whether an overview appeared, who was linked or whether your name was in the text. A monthly manual log of a fixed set of patient searches fills that gap, and it costs nothing but time.

Example: AI Overviews at a health practice

This example is illustrative, not a transcript. A functional medicine clinic in suburban Denver treats thyroid conditions. A patient searches "what helps with low thyroid symptoms." An AI Overview appears. It explains common causes, lists lifestyle and medical options, and suggests testing. Beside it are four links. Two go to national health sites, one to a hospital system, and one to a small naturopathic practice's article titled "Hypothyroidism: Which Labs to Ask For."

The Denver clinic ranks fifth in the classic results for "functional medicine Denver," and that is where its monitoring stops. It does not see that on this query, a smaller practice with a specific, useful article earned a link in the overview.

Look at what that article did. It targeted a subtopic, which labs to request, that fan-out would plausibly generate when expanding a thyroid question. It was indexed and readable. It answered in its opening lines. It had a named clinician. It met Google's stated bar, and the result was an AI citation earned with ordinary good practice.

Here is what the Denver clinic can do with that observation. Add a section to its thyroid page on the specific labs a patient should ask about, written plainly and attributed to a named clinician. Cover the neighboring questions, such as what normal ranges mean and when to retest. Keep the page indexed and snippet-eligible. Then rerun the thyroid searches monthly and note whether the clinic's pages appear among the links.

Notice what the clinic should not do. It should not add a special tag or file, because Google states none is needed. It should not rewrite its homepage in the hope of triggering an overview, because Google decides when one appears. And it should not report the result as a promise to leadership. No one can promise a place in a given overview. The goal is to raise the odds by being the clearest page on the subtopics patients actually ask about.

How to check yours in 60 seconds

1. Search Google for your top service plus your city, in a private browser window.
2. Search the three questions patients ask most before booking.
3. When a summary appears, look at the links beside it and note whether your site is one of them.
4. Write down which sites are linked instead, and open the first one.

Not every search shows an overview, so a blank result is normal. The wording of a summary often changes when you refresh it, while some main points and linked sites may repeat. I log what repeats and treat the rest as noise. Refresh once or twice and write down what repeats. Your own account also shapes what you see, so treat your result as one view (see AI personalization).

What to do about it

1. Confirm your key pages are indexed in Google Search Console.
2. Make sure no setting such as nosnippet or noindex blocks those pages from showing.
3. Put a direct answer in the first two sentences of each service and condition page.
4. Cover the sub-questions around a topic on the same page, such as tests, options, timelines, who it suits, and the cost and "versus" questions patients ask before booking.
5. Name a clinician and credentials on each page, and keep health claims within what you can support.
6. Record the same fifteen searches monthly, noting when an overview appears and who is linked.