How to get your business cited by AI search

For twenty years the goal of search was simple to state: be the first result. That was the whole game, and everything in SEO pointed at it. AI summaries have quietly broken the link between being first and being read.

This is not a prediction. There is now real measurement of what happens when an AI summary appears above the results, and of which pages those summaries cite. Both findings are uncomfortable if your plan is simply to rank.

What the click data shows

The most rigorous number available comes from Pew Research Center, which did not run a survey: it watched what people actually did. 900 US adults agreed to share their browsing, producing 68,879 real Google searches across March 2025.

• When an AI summary appeared, users clicked a search result on 8% of visits.
• When no summary appeared, they clicked on 15%, nearly twice as often.
• Users clicked a link inside the summary itself on just 1% of visits.
• 26% of visits with a summary ended the browsing session entirely, against 16% without.

Read the last two together, because that is the part most summaries of this study skip. The citation links inside an AI answer are not a meaningful traffic channel: they are clicked one time in a hundred. What a citation buys you is not a click. It is being the source the answer was built from.

Ranking first no longer means being cited

The assumption that carried over from classic SEO was that AI Overviews would simply summarise the top results. For a while that was broadly true. It is not any more.

Ahrefs analysed 863,000 keyword SERPs and 4 million AI Overview URLs using January 2026 data. Ahrefs sells SEO tools, so read it with that in mind, but this is a data study with its sample stated, not a case study about one of their customers. Only about 38% of the pages cited in AI Overviews also ranked in the top 10 for the query that triggered them. In their July 2025 study, that figure was roughly 76%. It halved in six months.

The explanation they give is query fan-out. Instead of answering your search directly, the system breaks it into related sub-questions, runs those, and assembles an answer from what it finds. The citations come from those sub-query results, which means a page can be cited for a question nobody typed, and the page ranking first for the actual search can be passed over completely.

What this changes about how you write pages

Cover the questions around the question

If citations are drawn from sub-queries, then a single page optimised for one head term is a thin target. The businesses that get cited are the ones with real pages answering the adjacent questions: what it costs, how long it takes, what happens if it goes wrong, how it compares to the alternative, whether it applies to their specific situation.

This is the same instinct behind a good FAQ, applied at page level rather than accordion level. Each substantial question a customer asks deserves somewhere it is answered properly.

Answer in liftable passages

A model assembling an answer needs a passage it can take whole. That favours writing where the direct answer comes first and the elaboration follows: a heading that asks the question, then a short paragraph that answers it without requiring the three paragraphs above it for context.

It is the opposite of the build-up-then-reveal structure that reads well in an essay. Put the answer first. You can still argue it afterwards.

Be specific enough to be worth quoting

Generic claims are not citable, because every competitor makes the same one. "Fast turnaround" is not a fact. "Most projects go from kickoff to launch in seven to ten days" is. Numbers, timeframes, named constraints and honest limits are what give a model something concrete to attribute to you.

Stay crawlable and stay accurate

None of the above matters if the page cannot be read or cannot be trusted. Content that only appears after JavaScript runs, pages blocked in robots.txt, and details that contradict what your Google Business Profile says are all ways to be excluded before the interesting part starts. The unglamorous technical work still comes first: see why your business might not be showing up on Google.

What this does not change

It would be easy to read the numbers above and conclude that search is over. That is not what they say. 18% of the searches in the Pew study produced an AI summary at all, which means the large majority still behaved like ordinary search results. Traditional rankings are still where most of the clicks are.

The shift is narrower than the headlines: for the informational questions that open a buying process, the answer increasingly arrives before the click, and you want to be inside that answer. For the transactional searches that end one (someone looking for a specific service near them, ready to call), the map pack and the organic results still do the work they always did.

How to tell whether it is working

Ranking reports will not show you this, and neither will a traffic chart, because the whole point is that the citation often does not produce a click. What you can track:

1. 01Ask the AI tools your customers use the questions your customers ask, and record whether you appear. Do it on a schedule, not once.
2. 02Watch branded search volume. If more people are arriving by typing your name, something upstream is introducing you.
3. 03Watch direct traffic and lead quality together. A caller who already knows what you do and what it costs has usually read about you somewhere you did not get a click from.

It is a less satisfying measurement story than a rankings chart, and anyone who offers you a clean number for it is overselling. The change is real, the direction is clear, and the work (depth, specificity, direct answers, clean technical foundations) is work that pays off in ordinary search regardless.

If you want to know where your site currently stands on the technical side of this, that is what our SEO work starts with: an audit that tells you what is wrong before anyone proposes writing anything.