How to Optimise for AI Overviews (What Actually Works)

To optimise for AI Overviews, do six things: rank the page in Google’s traditional results, open every section with a direct answer, make each H2 section self-contained, use tables and structured data, keep your brand entity unambiguous, and update the page honestly and visibly. That is the whole method. The rest of this guide covers how to optimise for AI Overviews in practice: the 2026 data on who gets cited (it changed sharply in March), the checklist we work through on client pages, before and after passage rewrites, and the honest unknowns most guides skip. We are Luckywebs, a London SEO and web agency, and this is how we approach it.

Disclosure: some links on this page are affiliate links. If you buy through one we may earn a commission at no extra cost to you. We only recommend tools we would genuinely use.

What is an AI Overview?

An AI Overview is the AI-generated summary Google places at the top of some search results pages, above the traditional organic listings. It is built by Google’s Gemini models, which read a set of source pages, write a short synthesised answer, and cite those sources as clickable links. AI Overviews launched in the United States in May 2024 and reached the UK in August 2024. They are not the same thing as AI Mode, which is a separate conversational tab in Google Search (more on that below). For website owners, the practical point is this: an AI Overview answers the searcher’s question directly on the results page, so the only way your site benefits is by being one of the cited sources. AI Overview SEO is therefore mostly about earning citations, not just positions.

How often do AI Overviews appear in 2026?

AI Overviews appeared on nearly half of the queries BrightEdge tracked in February 2026, with coverage up 58% on February 2025. Coverage is heavily skewed by intent: BrightEdge’s industry data shows healthcare, education and B2B technology queries triggering AI Overviews far more often than transactional or navigational searches. For contrast, Semrush’s study of 10 million keywords puts the November 2025 figure at just 15.69%: the gap comes down to which keyword sets each study tracks, both can be true at once, and neither is UK-specific. Either way, a large share of searches still shows no AI Overview at all, so classic organic rankings remain the whole experience for much of search. Here are the headline studies, with dates, because this data goes stale fast:

Source	Published	Finding (as reported)
BrightEdge	February 2026	AI Overviews trigger on nearly half of tracked queries, coverage up 58% on February 2025
Semrush	Data to November 2025	AI Overviews on 15.69% of 10 million studied keywords, after a July 2025 peak of 24.61%
Ahrefs	July 2025	76% of AI Overview citations came from pages ranking in the top 10
Ahrefs (update)	March 2026	38% of citations now come from top-10 pages; 31% rank 11 to 100; 31% outside the top 100

Who gets cited in Google AI Overviews?

The answer changed dramatically between 2025 and 2026, and most guides have not caught up. Ahrefs’ July 2025 study of 1.9 million citations found 76% of cited pages ranked in the top 10 for the same query. Its March 2026 follow-up, covering 863,000 SERPs and 4 million AI Overview URLs, found only 37.9% of citations now come from top-10 pages, with the rest split almost evenly between positions 11 to 100 (31.2%) and pages outside the top 100 entirely (31%). Ahrefs’ suggested explanation (a theory, not a confirmed cause) is query fan-out: Google splits a search into related sub-queries and cites pages that perform well for those, something its newer Gemini models appear to do more aggressively. The same update notes YouTube is now the most cited domain, up 34% in six months. Our reading: to rank in Google AI Overviews reliably, top-10 rankings are still the single strongest lever, but well-structured passages on lower-ranking pages now have a genuine route in that did not exist a year ago.

What about Google AI Mode in the UK?

Google AI Mode began rolling out in the UK on 28 July 2025, making the UK the third country to get it after the US and India, and notably without the Labs opt-in the other two markets went through first. AI Mode is a separate conversational tab in Google Search: you ask a question, get a full AI-written answer with linked sources, and can ask follow-ups. It differs from AI Overviews, which appear automatically inside normal results. For UK businesses the distinction matters less than the overlap: both systems synthesise answers from web pages and cite sources, and the same page-level optimisation work (answer-first passages, self-contained sections, clear entities) serves both. If your customers are in the UK, check how your brand appears in AI Mode UK results as well as in AI Overviews.

How do you optimise for AI Overviews? The working checklist

Optimising for AI Overviews means making each passage of a page quotable on its own, on a page Google already trusts. This is the sequence we work through on client pages:

1. Rank the page first. A top-10 position is still the strongest single predictor of citation. Do the ordinary SEO before the exotic SEO.
2. Open every section with the answer. Put a direct 40 to 60 word answer immediately under each heading, then expand. AI systems lift passages, not pages.
3. Make each H2 self-contained. Every section should name its subject and make sense quoted alone. No dangling pronouns.
4. Add structured formats. Comparison tables, numbered steps and definition passages are the formats AI Overviews quote most readily.
5. Use schema for clarity. Article, Organization, FAQPage and Person markup will not force a citation, but they remove ambiguity about who is saying what.
6. Nail entity clarity. One consistent brand name, a proper about page, author bios, and consistent NAP details everywhere your business is mentioned.
7. Keep the page fresh. Update the statistics, state the date you verified them, and change the copy honestly rather than bumping a date stamp.

Before publishing, we run this quick pre-flight:

• First 80 words answer the title’s question directly
• Every H2 reads as a question or task, answered in its own section
• At least one table or numbered list an AI could lift whole
• Claims dated and sourced, no orphaned statistics
• Schema validates and matches the visible content
• Author and organisation clearly identified

Before and after: two passage rewrites

Both examples below are generic illustrations written for this guide, not client copy; the pattern applies to any sector.

Example 1: the vague opener.

Before: “There are many factors to consider when choosing an accountant, and the right choice will depend on your individual circumstances.”

After: “A small UK business should choose an accountant on three criteria: sector experience, fixed monthly pricing, and support for Making Tax Digital software. Sector experience matters most, because an accountant who already serves your industry will know its allowable expenses and common HMRC pitfalls.”

Example 2: the dangling pronoun.

Before: “It also improves performance and is easier to maintain, which is why so many teams have adopted it in recent years.”

After: “Static site generation improves performance because every page is pre-built as plain HTML, so nothing is assembled at request time. Static sites are also easier to maintain, which is why frameworks like Astro have grown quickly.”

The rewrites are longer, but each one now survives being quoted alone: it names its subject, answers the implicit question, and gives a reason.

Does British spelling or the UK database change anything?

British spelling does not hurt your chances of citation: Google treats “optimise” and “optimize” as the same concept, and AI Overviews shown to UK searchers happily cite UK-spelled pages, so write naturally in British English. What does matter is measuring against the right market: track rankings and AI Overview presence on google.co.uk with UK location settings, because AI Overview coverage and citations differ by country. Check your keyword tool is reporting UK search volumes rather than US ones before you prioritise anything. On tooling, Semrush now bundles AI visibility tracking into its main plans: as of July 2026 the Starter plan includes 50 tracked AI prompts daily at $199 (about £146) a month billed monthly, or $165.17 (about £121) a month on annual billing (prices in USD, before VAT; see semrush.com/pricing). If Semrush is outside budget, our Semrush alternatives comparison covers cheaper routes to the same data.

How do you track AI Overview citations?

Google Search Console does not report AI Overview performance separately: clicks and impressions from AI Overviews are folded into the overall Search performance report, so you cannot isolate them there. That leaves three practical options, and we costed the paid ones in our comparison of AI visibility tools. First, manual spot checks on your priority queries from a UK location, screenshotting citations as you find them. Second, a rank tracker that records AI Overview presence and citations for your tracked keywords; we compared the options in our guide to GEO rank tracking, and our SE Ranking review covers the tracker we use most on client accounts. Third, watching referral patterns and branded search volume for indirect movement. For the wider toolkit question (which platforms cover classic rankings plus AI visibility without paying twice), start with our best SEO tools guide.

Is AI Overview accurate?

AI Overviews are usually accurate on well-documented topics and still make mistakes on ambiguous or thinly covered ones, which is why Google displays a disclaimer that AI responses may include mistakes. The early public failures (the 2024 examples that suggested glue on pizza) came from the system treating joke content as fact, and Google has tightened things considerably since. But accuracy is not a solved problem, and for site owners it cuts two ways. If an AI Overview quotes your content, unambiguous self-contained passages are harder to misread. If it describes your brand, check what it says: monitoring how AI systems characterise your business is now routine brand protection, not paranoia. A quarterly check of what AI Overviews and AI Mode say about your brand name, your services and your prices takes minutes, and it catches problems while they are still small.

What do we still not know?

An honest guide has to include this section, because AI Overview optimisation is built on correlation studies, not documentation. Nobody outside Google knows the actual selection mechanics; every study, including the Ahrefs and BrightEdge work cited above, observes patterns rather than rules. Nobody can promise inclusion: we can raise the probability of being cited, and no agency can guarantee it. The ground also moves fast: the fall from 76% to 38% of citations coming from top-10 pages happened in about eight months, apparently on the back of a single model upgrade, and the next upgrade could reshuffle things again. And some widely promoted tactics remain unproven: llms.txt, for example, is not used by Google at all, as we explain in our llms.txt guide. Treat anyone selling certainty here with suspicion.

Where does this fit into a wider GEO strategy?

AI Overview optimisation is one slice of generative engine optimisation (GEO): the broader work of earning visibility in ChatGPT, Perplexity, AI Mode and Google’s AI Overviews together. The good news is the work compounds, because the same well-structured, well-sourced, entity-clear pages perform across all of these systems, and the pages that get cited by AI engines tend to be the ones that already convert human readers. In practice that means you rarely need a separate AI content plan: you need one strong page per question your customers actually ask, structured so both people and machines can quote it, and a way of measuring whether the engines do. If you want a second pair of eyes on where your site stands, our GEO service starts with a review of how AI systems currently see and cite your brand.

Book Your Free AI Visibility Review

FAQ: optimising for AI Overviews

What is the difference between AI Overviews and AI Mode?

AI Overviews are automatic summaries at the top of normal Google results for some queries. AI Mode is a separate conversational tab that answers in full AI-written responses with follow-ups. Both cite web sources, and the same optimisation work serves both.

Can you opt out of AI Overviews?

Partially. Google’s snippet controls (nosnippet, data-nosnippet and max-snippet) limit how your content is used in AI Overviews, but they restrict your normal search snippets too, so opting out usually costs more visibility than it protects. No control blocks AI Overviews while leaving classic snippets untouched.

Does schema markup guarantee a citation?

No. Structured data clarifies what your content is and who published it, which helps AI systems use it confidently, but no schema type forces inclusion in an AI Overview. Treat schema as removing reasons to be skipped rather than adding a reason to be chosen.

How long does it take to appear in AI Overviews?

There is no fixed timeline, and anyone quoting one is guessing. Pages that already rank well can be cited soon after a restructure; new pages need to earn rankings first. We treat citations as a lagging indicator of the same authority work that drives ordinary rankings.

Do AI Overviews reduce clicks to websites?

Multiple independent studies have found lower click-through rates on queries where an AI Overview appears, which is exactly why citations matter: being the cited source captures a larger share of the clicks that remain, plus brand visibility inside the answer itself.

Is optimising for AI Overviews different from normal SEO?

Most of it is ordinary SEO done properly: rankings, crawlability, authority and genuinely useful content. The rest is passage-level discipline (answer-first sections that survive being quoted alone) and measurement, since standard reports do not show citations. Our guide to GEO rank tracking covers the measurement side.

Sources: Ahrefs, July 2025 and Ahrefs, March 2026 for citation data; BrightEdge data reported by Search Engine Journal, March 2026 for coverage data; Semrush, data to November 2025 for the contrast figure; Search Engine Land, July 2025 for the AI Mode UK launch. All figures checked July 2026.