Answer Engine Optimization: How to Get Your Business Cited by ChatGPT and AI Overviews

For twenty years the job was simple to describe: rank in the ten blue links. That job is changing underneath us. A growing share of the questions your customers ask never produce a list of links at all — they produce an answer, assembled by a model, with two or three sources cited underneath.

If you are one of those sources, you get the click and the credibility. If you are not, you are invisible in that conversation no matter how well you rank in the traditional results.

This is what people mean by Answer Engine Optimization (AEO), or Generative Engine Optimization (GEO). The names are new and slightly overlapping. The underlying discipline is not as unfamiliar as the acronyms make it sound.

The numbers that make this urgent

It is worth being specific about the scale, because "AI is changing search" has been said so often that it has stopped landing.

• Google AI Overviews now appear on an estimated 30–40% of all queries, answering directly without requiring a click.
• ChatGPT passed roughly 900 million weekly active users by February 2026 — more than double the year before — and handles over two billion queries a day.
• ChatGPT alone drives about 87% of all AI referral traffic to websites.
• AI-referred sessions grew over 500% year on year through 2025.
• Semrush projects that AI search visitors will overtake traditional organic visitors by 2028.

Two things follow from this. First, the traffic that does arrive from AI referrals tends to convert well, because the visitor has already been told by a trusted intermediary that you are a credible answer. Second, the traffic that does not arrive is invisible in your analytics — you cannot see the queries where a model recommended someone else.

The uncomfortable part of AEO is that failure is silent. A lost ranking shows up in Search Console. A lost citation shows up nowhere.

How answer engines actually choose sources

Models do not "crawl and rank" in the way a search engine does. Broadly, a query triggers a retrieval step that pulls candidate passages, and a generation step that composes an answer from them and attributes what it used.

That distinction matters enormously in practice, because it means the unit of selection is the passage, not the page. A 3,000-word guide that buries the answer in paragraph 40 competes badly against a page that answers the same question cleanly in 60 words under a matching heading.

What consistently gets picked up

1. A direct answer immediately after the question. If a heading asks "How much does a website cost in 2026?", the next sentence should answer it — not set up the answer.
2. Specific, checkable facts. Numbers, dates, prices, named tools and versions. Models favour passages that carry verifiable detail over passages that carry adjectives.
3. Clean structure. Real <h2> and <h3> elements, real lists, real tables. Headings styled with a <div> and a font-size are invisible as structure.
4. Original material. Your own data, your own pricing, your own case results. Synthesised summaries of what everyone else already wrote have nothing to cite.
5. Corroboration elsewhere. Being described consistently across directories, profiles and third-party mentions makes a model more confident in asserting a claim about you.

The practical checklist

1. Answer the question in the first two sentences

Take your most commercially important pages and check each heading. Does the text directly beneath it answer the question the heading poses, in one or two self-contained sentences? A passage that only makes sense with the four paragraphs above it cannot be lifted, and so will not be.

2. Publish a real FAQ block, in real markup

Not an accordion built from <div>s and JavaScript that only populates on click. Use <details>/<summary> or plain headings, and mark it up with FAQPage structured data. Retrieval systems handle question-shaped content well because the question text matches the query almost verbatim.

3. Get your structured data right

At minimum: Organization, WebSite, BreadcrumbList, plus Article on posts, Service on service pages, FAQPage where applicable and LocalBusiness if you serve a geography. Schema does not force a citation, but it removes ambiguity about who you are and what you sell.

4. Make the facts about your business consistent everywhere

Name, service list, locations served, founding year, contact details. If your site says one thing and your directory listings say another, a model resolving the contradiction may simply choose a competitor it can describe with more confidence.

5. Stop blocking the crawlers you want

Check robots.txt and any bot rules at your CDN. GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended each control different things, and they are not interchangeable. Blocking training crawlers while allowing search crawlers is a legitimate strategy — blocking everything by accident, via an over-eager security rule, is not.

6. Render your content server-side

Retrieval systems are far less patient than Googlebot with client-rendered content. If your key text only exists after a JavaScript bundle executes, assume some of them will never see it. Server-render the content that matters. This is one of several reasons we build marketing sites the way we do on our web development projects.

7. Track it, or you are guessing

Segment referrals from chatgpt.com, perplexity.ai, claude.ai, copilot.microsoft.com and gemini.google.com in your analytics. Then periodically ask the assistants themselves the ten questions that matter most to your business, and record who gets cited. It is crude, but it is the only direct measurement available.

What has not changed

It would be a mistake to read all of this as "SEO is dead, start again". Almost every input to AEO is something good technical SEO already asked for:

• Crawlable, server-rendered HTML
• Sensible heading hierarchy and semantic markup
• Fast pages — retrieval budgets are finite, and slow pages get sampled less
• Accurate structured data
• Genuine subject authority rather than keyword coverage

If your technical SEO is in order and your Core Web Vitals pass, you are most of the way there. The new work is mainly editorial: writing so that individual passages stand alone.

The mistake we see most often

Businesses respond to AI search by publishing more. More posts, longer posts, faster. This is precisely backwards.

Answer engines are not short of text — they are short of trustworthy, specific, attributable text. Twenty thin articles assembled from what already ranks give a model nothing it cannot get elsewhere. One page containing your real pricing, your real methodology and your real results is citable, because nobody else has it.

Publish less. Publish what only you can say. Structure it so a machine can lift the answer cleanly.

Where to start this week

1. List the ten questions a buyer asks before hiring in your category.
2. Ask ChatGPT, Perplexity and Google each one. Record who gets cited.
3. For every question where you are absent, check whether you have a page that answers it directly under a matching heading. Usually you do not.
4. Write those pages — short, specific, factual, structured.
5. Re-check in six weeks.

That loop is the whole discipline. Everything else is refinement.