How to rank in AI Overviews: the page changes that matter

7 min read

How to rank in AI Overviews: the page changes that matter

7 min read

A page that ranks third for a valuable search can still be missing from the AI Overview above it, while a competitor on page three is cited. Teams asking how to rank in AI Overviews often assume the answer is to rank higher in the blue links. The data says otherwise.

In March 2026 Ahrefs found that 38% of pages cited in AI Overviews also ranked in the top 10 for the same search, down from about 76% in its July 2025 study. This article covers what we change on the page to get cited, why it works, and how to tell whether it is working.

Why AI Overview citations matter to the business

An AI Overview takes a single position at the top of the results, and every link inside it shares that position, according to Google's Search Console documentation. If your brand is cited there, it is in front of the buyer before the organic listings. If a competitor is cited instead, theirs is the name the buyer reads first.

Google says clicks from results pages with AI Overviews are higher quality, with people more likely to spend longer on the site. Google has not published figures for this, so treat it as a claim to test in your own analytics. What is clear is that a citation can reach a buyer on searches where your page is not in the top 10 at all.

AI Mode adds to the stakes. Google says it gives a full AI response with links to supporting websites, and Search Console counts each follow-up question in AI Mode as a new query. Each follow-up is another chance to be cited, or to be replaced.

Getting it wrong costs quietly. Rankings can hold while the answer above them names someone else, and no rank report will show it.

How we change a page to get cited in AI Overviews

Google's guidance on AI features says there are no extra technical requirements and no special schema to add. What changes is how the page answers the question and the questions around it. These are the steps we follow.

We start with pages that already rank on page one or two for searches that show an AI Overview, because Google already treats them as relevant. New pages come later, once we know which sub-questions nobody answers well.

Confirm the page is eligible. Google says a page must be indexed and eligible to show with a snippet. We check for nosnippet, max-snippet and data-nosnippet rules, and that robots.txt and the CDN allow Googlebot. Google-Extended is a separate control for Gemini training and grounding and, Google says, does not affect inclusion in Search.
Map the query fan-out. Google says AI Overviews and AI Mode may use query fan-out, issuing multiple related searches across subtopics and data sources. We list the sub-questions behind the main search using People Also Ask, related searches, AI Mode follow-up questions and the sources the current AI Overview cites. That list becomes the brief.
Put a direct answer near the top. Under the H1 or first heading, we write a self-contained answer of two or three sentences that would make sense if quoted alone: it names the subject, gives the answer and adds the main condition. Introductions that take three paragraphs to reach the point are one of the patterns we change most often.
Cover the sub-questions on the same page. Each sub-question from the fan-out gets a heading in plain question form and an opening sentence that answers it. This is the likely reason cited pages so often sit outside the top 10, and the one Ahrefs points to: a page can be the best answer to a sub-query even when it is not the best answer to the main search.
Name sources, add numbers and quote people. We attribute every fact to a named, linked source, add verified statistics and include quotations from named experts where they add something. The GEO paper by Aggarwal and colleagues, accepted at KDD 2024, found that citing sources, adding quotations and adding statistics improved visibility in generative answers by 30% to 40% on its main measure, while keyword stuffing gave little or no improvement.
Make the content easy to extract. Comparisons go in tables, processes in numbered lists and definitions in one sentence. Important content must be in text rather than images, and structured data must match the visible text: both are on Google's list of best practices for AI features. Our guide to writing content that converts and ranks covers the wider page.
Make it findable and clearly owned. We link to the page from related pages so Google can find it, add a named author with a short biography, and keep the date and facts current. A page that states who wrote it and why they know the subject gives a reader, and a system choosing sources, a reason to trust it.
Apply the same work for Perplexity. Perplexity says PerplexityBot is the crawler that surfaces and links sites in its results, so we check it is allowed in robots.txt and that its published IP ranges are not blocked. The GEO paper tested Perplexity directly and reported visibility gains of up to 37% there from the same kind of changes.

What a rewritten page looks like

Take a guide aimed at a search such as how long a CRM implementation takes. The original opens with two paragraphs on why CRM matters and gives the answer halfway down. The rewrite answers in the first two sentences: a typical range, and the three factors that move it.

Below that, each fan-out sub-question gets its own heading: what affects the timeline, how data migration changes it, what to do in the first 30 days, and how timelines differ between vendors. Each section opens with its answer, then gives the detail, and every figure links to where it came from.

Nothing commercial is removed. The product comparison, the case for booking a demo and the internal links stay where they were. The page becomes easier to quote, and easier to read.

How to measure AI Overview citations

We track three things. First, citation presence: a fixed set of priority searches checked weekly, recording whether an AI Overview appears and which domains it cites. Second, Search Console's generative AI performance report, available to all sites since 31 August 2026, which shows impressions in AI features by page, country, device and date, but not clicks or queries. Third, GA4 landing page data for the changed pages, compared with the period before.

Good looks like a rising share of priority searches where you are cited, and rising AI feature impressions on the pages you changed. We report it to leadership as the share of priority searches won against named competitors, which is easier to act on than a raw count. To see which sources currently shape the answers you want to appear in, this free prompt is a quick place to start.

Be clear about the limits. AI Overviews do not appear for every search and can change between checks, and Search Console does not report clicks from them separately. We allow six to eight weeks after a change before judging it.

When it does not fit, and the mistakes we see most

If your priority searches rarely show an AI Overview, this work comes after standard SEO on those pages. Check the searches before you rewrite anything.

Rewriting pages for the AI Overview and losing what made them convert. Keep the commercial content and add the answers.
Adding a long FAQ block of questions nobody asks. Use the fan-out list, not invented questions.
Expecting schema to earn the citation. Google says no special markup is needed.
Repeating the keyword. The GEO paper found keyword stuffing gave little or no improvement.
Judging from one screenshot. Check the same searches on a schedule.
Quoting statistics you cannot source. It is a risk to the brand as well as the page.
Changing every priority page at once, which makes it impossible to tell which change worked.

Questions to ask your team or agency

Our AI SEO and GEO team runs this page work alongside citation tracking. Whoever does it, these questions show whether it is on track.

Which of our priority searches show an AI Overview, and who is cited in them?
What sub-questions does each priority page answer, and which are missing?
Where on the page is the direct answer, and could it be quoted on its own?
Which facts on our key pages have a named, linked source?
How are we tracking citations from week to week?

The same page changes sit behind our answer engine optimisation work, which structures content so assistants and featured snippets can lift a direct answer.

Frequently asked questions

How do you rank in AI Overviews?

There is no separate ranking system to target. Google says a page needs to be indexed and eligible for a snippet, with no extra technical requirements or special schema. Pages tend to be cited when they answer the main question directly near the top, cover the sub-questions Google's query fan-out searches for, and back their claims with named sources and figures.

Do you need to rank in the top 10 to appear in AI Overviews?

No. In a March 2026 study of 863,000 keyword results, Ahrefs found that 38% of pages cited in AI Overviews also ranked in the top 10, down from about 76% in July 2025. Google's AI features issue several related searches across subtopics, so a page can be cited for answering a sub-question even when it ranks lower for the main search.

How do you rank in Perplexity?

Make sure PerplexityBot is allowed in robots.txt and not blocked by your firewall, since Perplexity says that crawler is what surfaces sites in its results. Then write pages it can quote: a direct answer near the top, clear headings for related questions and facts tied to named sources. Research that tested Perplexity found visibility gains of up to 37% from adding citations, quotations and statistics.