How to rank in Google AI Overviews

Google's AI Overviews now sit above the normal results for a large and growing share of searches, summarising the answer and citing a handful of sources. For many informational queries, that summary is the first thing your buyer reads, and sometimes the only thing. So the question marketing leaders keep asking is fair: how do you get your page into the Overview rather than buried beneath it. The honest framing is that you do not rank in an Overview the way you rank in the blue links, and understanding why changes what you should actually do.

You do not rank in an AI Overview directly. Google composes it from pages that are already performing in its index for the query, so the route in is to earn strong conventional visibility for the query and its sub-questions first, then make your answer easy to extract: a direct response near the top of the page, passages that stand on their own, and coverage of the related questions the Overview breaks the query into. There is no separate Overview index to submit to and no placement to buy. This article explains the mechanism and the work that follows from it.

How an AI Overview is actually built

Start with the mechanism, because most bad advice about Overviews comes from imagining them as a separate ranking system rather than a layer on top of the one you already know.

When a query triggers an Overview, Google does not compose it from nothing. It draws on its existing understanding of the query and its index of pages, and it uses a technique often described as query fan-out: it breaks your search into a set of related sub-questions, retrieves pages relevant to each, and synthesises the answer from across them, citing the sources it leaned on. This is why the pages cited in an Overview are usually pages that already rank well for the query or its components. The Overview is built on the back of conventional relevance, not in place of it.

Two things follow immediately. First, conventional visibility is the foundation: if your page is not in the running for the query and its sub-questions, it is not a candidate for the Overview either. Second, the unit of selection is a passage that answers a specific sub-question, not a whole page that is generally about the topic. A page can rank respectably and still be passed over for the Overview because it never states any single answer cleanly enough to be lifted. Both of these shape the work, and neither is exotic.

Earn the conventional visibility first

Because the Overview is drawn from pages that already perform, the first and largest piece of the work is ordinary, rigorous search visibility for the query and everything around it.

That means the fundamentals have to be genuinely in place: the page is crawlable and fast, its content is present in the server-rendered HTML rather than painted in by JavaScript a crawler will not run, and it is relevant and authoritative enough to rank for the query in the normal results. If you are not showing up in the top handful of conventional results for a query, chasing its Overview is premature, because you are trying to be selected from a candidate set you have not yet entered. The unglamorous truth is that most Overview visibility is won by being a strong conventional result, and the teams that treat Overviews as a brand-new discipline tend to skip the foundation that actually feeds them.

This is the same relationship that AI assistants have to search generally, and it is why "SEO is dead" is exactly backwards for this surface. The Overview did not replace the index. It sits on top of it and made being in it more valuable, because now a strong page can be cited in the summary as well as listed below it.

Cover the sub-questions, not just the headline query

Once the foundation is there, the fan-out mechanism tells you something specific about content: breadth of coverage matters, because the Overview is assembled from answers to the sub-questions the main query decomposes into.

Think about what someone really wants when they search your target query, and map the cluster of related questions around it, the definitions, the comparisons, the "how", the "how much", the objections. A page, or a small set of linked pages, that answers that whole cluster clearly gives Google more surfaces to cite than a page that answers only the headline question. This is not an argument for padding. It is an argument for genuine completeness: covering the adjacent questions a thoughtful reader would ask next, each answered plainly, so that whichever sub-question the fan-out lands on, you have a clean passage for it.

A useful way to find those sub-questions is to read the "People also ask" box and the related searches for your target query, and to notice which sub-questions the current Overview itself is answering and citing. Those are Google showing you, directly, the components it broke the query into. Answering them well, in extractable form, is much of the game.

Make the answer extractable

The second half of the work is structural, and it is the same discipline that drives citation everywhere in AI search: a passage a system can lift cleanly into a synthesised answer.

Open the section that addresses each question with a direct, self-contained answer, before the elaboration, so the response exists in a single liftable passage rather than being distributed across several paragraphs the reader has to assemble. Write those passages so they stand alone, making sense out of context rather than depending on a long run-up. Use headings that state what the section answers rather than tease it. Keep the language plain and specific, because a hedged, throat-clearing paragraph is a poor extraction candidate and a crisp one is a good one. Where it fits, a short, direct question-and-answer block is among the most extractable structures there is, which is why it earns its place near the foot of a page.

The reassuring part, again, is that none of this is writing for a machine at the expense of a person. A reader skimming for the answer wants exactly what the extraction step wants: the point, stated early and clearly. The mechanics of why extractable structure drives citation, across Overviews and assistants alike, are set out in how AI assistants decide which sources to cite.

The measurement problem, and how to work around it

Here is the catch that makes Overviews harder to manage than ordinary rankings: you cannot measure your presence in them cleanly from your own analytics.

Google Search Console does not break out AI Overview impressions or clicks as a separate line. There is no report that tells you "you appeared in this many Overviews and earned this many clicks from them". Overview impressions are folded into your general Search performance, which means you cannot open a dashboard and read your Overview visibility off it. You have to observe it directly instead. Take your fixed set of target queries, run them as ordinary Google searches, and record whether an Overview appears, whether it names or cites you, and which other sources it draws on. Do this on a schedule with the same queries each time, so you can see change rather than a single snapshot.

The traffic effect has to be inferred rather than read directly, and the tell-tale pattern is a page holding its ranking position while its clicks decline, which usually means an Overview is answering the query before the user reaches your link. I have set out how to diagnose that pattern, and separate it from a genuine ranking loss, in why your organic traffic is falling while your rankings hold. For now, the practical point is that Overview visibility is something you watch by hand, query by query, because the platform does not hand it to you.

Whether you even want in, on every query

It is worth pausing on a strategic question most Overview advice skips: appearing in an Overview is not unambiguously good on every query, and treating it as a universal goal can cost you.

On a query where the Overview fully answers the question, being cited in it keeps your brand visible but may earn you fewer clicks than a strong conventional result once did, because the user got what they needed on the results page. On a query where the answer is genuinely complex, or where the buyer needs to evaluate and choose, the Overview tends to send the motivated user onward, and a citation there is close to pure gain. So the sharper way to think about it is not "how do I appear in every Overview" but "for which queries is an Overview citation worth pursuing, and for which is my effort better spent on the deeper, decision-stage pages that Overviews cannot replace". The decision-stage content your buyers read before they choose is rarely reducible to a summary, which is precisely why it keeps earning the click.

You can technically keep a page out of Overviews using the nosnippet directive, but it is a blunt instrument: it also removes your ordinary snippet, costing you conventional visibility to avoid the Overview. For almost everyone, the answer is not to opt out but to be a strong, extractable source and to spend your deeper effort where summaries cannot compete.

Why a strong page still misses the Overview

Even a page that ranks well and reads clearly can be left out of the Overview, and the reasons are worth knowing because each one points at a fixable gap rather than bad luck.

The most common is that the page answers the headline query but not the specific sub-question the fan-out landed on. Google decomposed the search into components, retrieved a passage for each, and yours addressed the topic in general while a competitor addressed the exact component directly. The fix is coverage: answer the adjacent sub-questions explicitly, each in its own clearly headed passage, so there is a matching answer whichever component the Overview assembles from.

The second is that the answer is present but not liftable. The information is on the page, distributed across a few paragraphs the reader has to synthesise, rather than stated in a single self-contained passage. A model composing a summary reaches for the source that already did the summarising, so a page that makes the reader assemble the answer loses to one that states it outright.

The third is freshness. Overviews lean toward content that looks current for queries where recency matters, so a page not reviewed in two years can be passed over for a newer one saying much the same thing, even when yours is better. A genuine, visible update, not a changed date, can bring an older strong page back into contention.

And the fourth is trust on the specific claim. For queries where accuracy carries risk, Google is more conservative about which sources it will synthesise, favouring those with clear authorship, corroboration, and a track record on the topic. On those queries the missing ingredient is often not structure but standing, which is slower to build and harder to fake. The way to read all four reasons is as a diagnosis: when a strong page misses the Overview, one of them is almost always behind it, and each points at specific work to do next rather than leaving you to guess or to conclude that Overviews are simply arbitrary.

The takeaway

You do not rank in an AI Overview on its own terms. You earn strong conventional visibility for the query and the sub-questions it fans out into, then structure your answers so they can be extracted cleanly, then observe your presence directly because Search Console will not report it for you. The foundation is ordinary search rigour; the addition is breadth of coverage and answer-first structure; the discipline is measuring by hand on a fixed query set.

If you want a measured read of where you currently appear across AI Overviews and the assistants your buyers use, and a prioritised plan for the queries worth pursuing, that is what an AI search visibility audit provides.

This article is part of the SEO in the AI Era: The Complete Guide guide.