What is AEO? Answer engine optimization, explained for small businesses

A growing share of your customers no longer see ten blue links when they search. They ask ChatGPT, or Google answers them directly with an AI summary, and the answer names two or three sources — or none. Answer engine optimization, AEO, is the work of being one of the sources the answer names. You'll also see the term GEO, generative engine optimization; the difference between the two matters to consultants selling them and to almost nobody else.

This is worth your attention, but not your panic. Most of what earns AI citations is the same work that earns Google rankings, and most of what's sold as special AEO technique is either repackaged SEO or folklore. Here's what the actual evidence says.

What changed, in numbers

Pew Research tracked the real browsing of 900 US adults across 68,879 searches in 2025. When a Google search produced an AI summary, people clicked a regular result 8% of the time, versus 15% when there was no summary — roughly half the clicks gone. Links inside the AI summary itself were clicked on just 1% of visits.

So fewer clicks overall, and the ones that remain concentrate on whoever the answer cites. That's the whole strategic picture: the pie shrank, and the slices go to named sources. Being unnamed in an AI answer is the new page two.

The good news: it's mostly still SEO

Google's own documentation on AI features says there are no additional requirements to appear in AI Overviews and no special optimizations necessary — the same SEO best practices apply. That matches how these systems work: AI answers are assembled from content that ranks well in the underlying search index. If your pages are indexed, rank for the question, and answer it clearly, you're already in the pool the AI draws from.

Which means the fundamentals you'd do anyway — topics matched to what buyers actually search, pages that answer one question well, a site Google can crawl — are also the entry fee for AI visibility. There's no separate game. There's the same game with a new scoreboard.

What actually increases citations: the Princeton findings

The best evidence on tactics comes from the 2023 study that coined the term GEO — researchers from Princeton, Georgia Tech, the Allen Institute and IIT Delhi tested nine content changes across a 10,000-query benchmark and measured which made generative engines cite a source more. Three stood out:

• Adding statistics — concrete numbers in the text improved visibility in AI answers by as much as 40%, the strongest single change tested.
• Adding quotations — direct quotes from named people or organizations produced a similar-sized lift.
• Citing sources — content that references where its claims come from got cited more itself.

And one finding worth framing: keyword stuffing — the classic bad-SEO move — made results slightly worse than doing nothing. AI systems quote content that sounds like evidence, and they skip content that sounds like advertising.

AI answers quote content that sounds like evidence and skip content that sounds like advertising. Write the page a careful journalist would cite.

Where AI answers get their sources (and why you shouldn't chase it)

Semrush studied which domains AI answers cite most, and two names dominated: Reddit, appearing in roughly 40% of answers, and Wikipedia at about 26%. But the study's real lesson was volatility — citation patterns swung sharply within weeks when models were updated, and each platform favors different sources. Whatever the citation pattern is today, it can be different by next quarter.

The practical takeaway for a small business isn't to chase the currently favored platform. It's that AI assistants read the whole public web about you — your site, review platforms, directories, forum threads where your category gets discussed. Being genuinely present in several of those places is robust to model updates. Being optimized for one is not.

What to skip

• llms.txt — a proposed file that tells AI systems about your site. Google has said it doesn't use it, and Google's John Mueller has compared it to the long-dead keywords meta tag, noting that AI services don't even request the file. Harmless, but zero evidence it does anything.
• AEO retainer add-ons without deliverables. Agencies have started billing AI search optimization as a separate line item. Given that the evidence says AEO is mostly good SEO plus better-evidenced writing, ask precisely what work that line buys before paying it.
• Anything promising guaranteed AI citations. The citation patterns shift with every model update; nobody controls them, same as rankings.

The small-business AEO checklist

Everything the evidence supports, in one list:

• Keep doing regular SEO — indexed, ranking, relevant. It's the qualifying round for AI visibility, and there are seven signals for checking it's on track.
• Put numbers in your important pages: prices, timelines, measurements, results. Statistics were the strongest citation driver in the Princeton study — and pages with concrete facts convert better anyway.
• Answer questions the way people ask them. A clear question as a heading, a direct answer in the first sentence under it, detail after. AI systems extract answers; make yours extractable.
• Keep your Google Business Profile, reviews, and key directories current — AI assistants recommending local businesses lean on the same public signals local search does.
• Watch for AI referrals in your analytics (traffic from chatgpt.com, perplexity.ai and friends). Small numbers today, but they tell you when this starts mattering for your niche.

None of that is a new discipline. It's the same bet as SEO: publish genuinely useful, verifiable answers to the questions your customers ask, and be findable wherever they ask them. The machines reading your site have changed. What they reward, so far, mostly hasn't.