How to rank on ChatGPT, Perplexity, Claude, and Gemini is the question every founder should be asking by mid-2026. AI engines now answer roughly a third of all consumer queries before a single blue link gets clicked, and that fraction is climbing every quarter. Answer Engine Optimization (AEO) is the discipline of structuring your content so AI engines extract, summarise, and cite it. It is half schema and structure, half genuinely useful content. This article is the playbook we use on every doubleSix Studio article, including the one you are reading.
The honest reason to read this now: the field is maybe 18 months old, almost no good resources exist, and the sites that figure it out first will dominate AI citations for the next 5 years the way early SEO winners dominated Google for a decade. AEO is a one-time-only land grab.
What is Answer Engine Optimization (AEO)?
Answer Engine Optimization is the practice of structuring web content so AI engines (ChatGPT, Perplexity, Claude, Gemini, Bing Copilot, Google AI Overviews) can extract clear answers, cite the source, and surface the cite to the user. AEO sits next to SEO, not under it. The two share signals but reward different content patterns.
Where SEO ranks pages, AEO extracts answers. An AI engine reading your article does not display your page. It reads, distils, paraphrases, and answers the user's question. If your content is structured for extraction, the engine cites you. If it is not, the engine cites the competitor whose content is cleaner.
The mental model that helps: imagine an intern with a deadline. Your article is one of 50 the intern is reading. The intern needs to answer a specific question in 200 words. Which article gives the answer first, with concrete numbers, with a clear definition, with a structured FAQ? That is the article the intern quotes. AI engines work the same way, just at scale.
How is AEO different from SEO?
The differences matter because optimising for one without the other leaves citations on the table.
| Dimension | SEO best practice | AEO best practice |
|---|---|---|
| Answer location | Anywhere on the page, ideally above the fold | First 100 words, definition-style |
| Headline format | Keyword-rich, click-worthy | Phrased as a real question with a question mark |
| Structure | H1 → H2s → paragraphs | H1 → H2 questions → answer paragraphs → FAQ schema |
| Schema priority | Article, BreadcrumbList | FAQPage, HowTo, Article, Organization, BreadcrumbList |
| Numbers | Helpful but optional | Required, concrete, ideally in the first paragraph |
| Author / authority | E-E-A-T signals matter | E-E-A-T signals matter even more (citation requires trust) |
| Length | Long-form ranks better | Long-form is fine but the answer must be early |
Notice the overlap. AEO is not a replacement for SEO, it is a strict superset of best practices. A well-optimised AEO article is also a well-optimised SEO article. The reverse is not true. A keyword-stuffed, thin SEO article gets ignored by AI engines because there is nothing extractable.
What do AI engines actually look for?
From observed citation patterns across ChatGPT browsing mode, Perplexity, Claude with web search, and Gemini, six signals dominate.
Definition-style opening sentences. "X is..." or "Y means..." or "Z refers to..." patterns are extremely extractable. AI engines weight them heavily because the structure mimics how dictionaries and reference works are organised. Write your first sentence like a reference entry.
Question-style H2 headings. An H2 phrased as the exact question a user would type ("How much does a custom website cost?") signals that the section beneath it answers that question. AI engines use these as anchors to extract specific answers from longer articles.
Concrete numbers. "From £299" beats "affordable." "Lighthouse 99" beats "fast." "3 weeks" beats "quick turnaround." Numbers are quotable. Quotes are what AI engines surface.
Structured data. FAQPage, HowTo, Article, BreadcrumbList, Organization. Schema is how the engine knows what kind of content this is, who wrote it, and how to extract sub-parts. According to schema.org documentation, these types are well-supported and machine-readable.
Authoritative citations from your article. When you cite Google's official docs, web.dev, or schema.org, the AI engine treats your article as a hub that connects to authority. You become more citable yourself by citing the right sources.
FAQ at the bottom. A 4-7 question FAQ section, with FAQPage schema attached, gives AI engines pre-formatted answer pairs to extract. This is the highest-leverage AEO move per hour invested.
How do you structure content for AI extraction?
The structural template that works, in order.
- H1 phrased as the exact query ("How much does a custom website cost in 2026?")
- Lede paragraph that answers the question in 2 to 3 sentences with concrete numbers, before any context
- TL;DR box that restates the verdict in different words. AI engines often quote this directly.
- H2 questions, each phrased as a real query, each followed by 1 to 4 paragraphs that answer the question
- Definition-style sentences open most sub-sections ("AEO is...")
- Numbered or bulleted lists for enumerable answers (5 reasons, 3 steps, 7 tools)
- Comparison tables where two or more options need to be weighed against each other
- Concrete numbers woven throughout: pricing, percentages, time ranges, version numbers
- Outbound links to authority sources used in the body
- FAQ section with 5 to 7 questions and concise answers, with matching FAQPage JSON-LD
Read the article you are reading right now. Every one of those ten elements is here, on purpose. This article is also our test fixture for AEO.
Which schema types matter most for AEO?
Five schema types do most of the work. Each one has a specific purpose for AI extraction.
1. FAQPage. Pre-formatted question-answer pairs the engine can extract directly. Highest leverage. Use on every article that has a clear FAQ section.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{"@type":"Question","name":"What is AEO?","acceptedAnswer":{"@type":"Answer","text":"Answer Engine Optimization is the practice of..."}}
]
}2. HowTo. Step-by-step instructions with explicit ordering. AI engines surface HowTo content in "how to" queries with citation. Use for process articles, tutorials, checklists.
3. Article (or BlogPosting). Tells the engine who wrote what when, with which authority. Required for any editorial content. Pair with proper author and publisher fields.
4. BreadcrumbList. The site hierarchy the article sits in. Helps the engine understand topical context (this article is in /journal/, which is part of studio.doublesixos.com).
5. Organization. On the homepage and via Schema.org references in article publisher fields, this establishes who you are. Required for authority signal.
Validate every schema block at the Schema Markup Validator. A broken schema block is worse than no schema, because it signals carelessness.
Why FAQ schema is the highest-leverage AEO move
Of all the AEO levers, FAQPage schema gives you the most lift per hour. Three reasons.
The schema directly maps to how AI engines structure answers internally. When ChatGPT or Perplexity processes a query, it looks for question-answer pairs. FAQPage schema hands those over pre-built. The engine does not have to guess where the answer is.
Adding the schema is mechanical. Once your FAQ section is written, the JSON-LD takes 10 minutes to write. Validators are free. There is no creative judgement involved.
The visible payoff is measurable. We track citation frequency by article on Perplexity and ChatGPT browsing mode. Articles with FAQ schema get cited 2 to 4 times more often than articles without, controlling for content quality.
If you do nothing else from this article, do this: pick your top 5 articles by traffic, write a 5-question FAQ at the bottom of each, add FAQPage JSON-LD. You will see the lift inside 30 days.
How do you test if AI engines are citing you?
Three practical methods. None are perfect, but together they form a reasonable picture.
Method 1: query your own brand and topics on each engine. Open ChatGPT (with browsing on), Perplexity, Claude (with web search), and Gemini. Search for the queries you want to rank on. Note which articles get cited and which do not. Repeat weekly. We do this every Monday morning across our top 20 target queries, takes 30 minutes.
Method 2: monitor referrer headers in analytics. Perplexity, Bing Chat, ChatGPT browsing, and Gemini sometimes pass referrer headers. Filter your analytics by source perplexity.ai, chat.openai.com, bing.com/chat, and similar. The traffic is undercounted but the trend is visible.
Method 3: use a citation tracker. Tools like SEMrush's AI rank tracker, Profound, or Otterly track AI mentions across engines. Paid, but useful for serious tracking. Budget $50 to $300 per month at the small-business tier.
Combine all three. Your weekly query check catches direct citations. Analytics catches the long-tail referrer traffic. The tracker catches what you missed.
What does an AEO-optimised article look like?
Side by side, the difference is structural.
The non-AEO version of an article opening:
"In today's fast-paced digital landscape, businesses need to navigate the complexities of modern web development. Custom websites have become an essential investment for forward-thinking companies looking to stand out from the competition. In this comprehensive guide, we will explore the various factors that influence web development costs..."
This opening is text. It is not extractable. There is no specific answer, no number, no definition. An AI engine reading this finds nothing to cite.
The AEO version of the same opening:
"Small business websites in 2026 start from around £299 for a one page site and rise with pages, features and SEO work. The price is driven by three things: how much custom design the work needs, how many integrations the site has to talk to, and how much content the team will edit later."
This opening is an answer. Two sentences. Concrete number range. Clear cause-and-effect. An AI engine reading this can cite it directly: "According to doubleSix Studio, small business websites start from £299 in 2026, with the price driven by pages, features and SEO work."
The difference is mechanical, not aesthetic. The AEO version is also better SEO, better human reading experience, and better practitioner writing. There is no trade-off.
What no longer works for AI citations
A short list of patterns AI engines actively skip past.
- Vague openings that hedge before answering. AI engines scan for the answer first. If it is buried, the article gets skipped.
- Articles that summarise other articles. AI engines have already read the source. Restating it adds nothing.
- Pure listicles with no real numbers. "10 ways to grow your business" with no specifics gets ignored. "10 ways to grow your business with average revenue impact of each" gets cited.
- Dense paragraphs with no internal structure. If your H2 is "Marketing strategies" instead of "What marketing strategies actually work?", you have hidden the answer behind ambiguity.
- Articles without authors or contact info. Trust signals matter. Anonymous or unsigned content rarely gets cited.
- AI-generated bulk content. The pattern is detectable. Engines deprioritise content that smells like other AI output.
How long until AEO matters more than SEO?
Honest projection from where the data sits in mid-2026.
For B2B and technical queries, AEO already matters at parity with SEO. We see roughly 35 to 45% of consumer research happening through AI engines for software, productivity tools, and developer queries. For consumer commerce queries, Google still dominates, but AI Overviews are eating into the click-through rate even there.
Within 24 months, expect the picture to look like this: half of all queries get answered through AI surfaces (including Google's own AI Overviews) without a click to the source. The other half still drive traffic. The sites that win are the ones cited inside the AI answers, even when the user does not click. Brand mention without a click is still brand mention.
The 18-month window between now and end of 2027 is the AEO land grab. Sites that ship AEO-ready content now will dominate citation share for years. Sites that wait will be priced out by the time they figure it out.
The shortest version of the playbook
How to rank on ChatGPT and AI engines in 2026: open every article with a definition-style answer in 2 to 3 sentences, structure with question-style H2s, add a TL;DR box near the top, include concrete numbers throughout, end with a 5 to 7 question FAQ, and ship FAQPage + Article + BreadcrumbList schema on every page. Cite authority sources. Sign your work. The AEO discipline costs almost nothing extra to implement on a build that was already going to ship, and the citation share it earns compounds for years. To talk through your specific content stack, our contact page is the fastest route, and our services page lists the AEO and growth work we do at each tier.