AEO is the practice of structuring content for extraction by Google's AI Overviews and featured snippets. The rules are concrete: direct answers in the first 60 words, FAQPage schema, bulleted lists, comparison tables, and question-shaped headers. Pages built this way get extracted within weeks. Pages that ignore the format don't get extracted at all, no matter how good the content is.
01 / Why does AEO matter more than it sounds?
AI Overviews now appear above the traditional search results on roughly half of informational queries on Google. When they appear, click-through to the #1 organic result drops 30 to 50 percent.
Ranking #1 in the traditional results is worth half what it used to be on those queries. The traffic didn't disappear. It got reassigned to whoever Google cites as the source. The new top slot isn't position one. It's position zero, which is the cited source above the traditional results in an AI Overview.
Position zero is what AEO wins.
Getting extracted is only useful if the product is worth finding. The Vibe Founder Trap covers the decision that comes before the distribution.
AEO isn't a separate skill set from SEO. It's the modern formatting layer on top of it. You still need the page to rank in the top 10 for the query before extraction is even possible. AEO turns a top-10 page into the cited source. Skip AEO and you're competing on the old game while everyone else is playing the new one.
If you've been doing SEO and ignoring AEO, you're getting outranked by lower-authority pages that just structured their content for extraction. That's the current state of search in 2026. On the pillar article at localhost3000.agency/notes/seo-aeo-geo-llmo, I tracked extraction patterns across 12 published pages over six weeks. The pages with proper AEO formatting were cited in AI Overviews at roughly 2.3x the rate of identical-content pages without it. The structure does the work.
02 / What does Google's AI extract from a page?
Three formats dominate AI Overview extraction.
Direct answers. Two to four sentences that resolve the query immediately, ideally in the first 100 words of the page. Definitional queries ("what is X") get extracted from the page's opening lines about 80 percent of the time. Burying the answer kills extraction.
Bulleted lists. AI Overviews preferentially extract lists because they're already structured data the model doesn't need to parse. If your content can be expressed as a list, make it a list. Five to seven items is the sweet spot. Lists under three items look thin. Lists over ten lose extraction priority.
Comparison tables. The single highest-extraction format of 2026. When a query implies a comparison ("X vs Y," "best X for Y"), Google's AI lifts entire tables into the Overview when they're well-formatted. The pillar article uses one comparison table and it's the single most extracted block of content in the piece.
What AI Overviews rarely extract:
- Long flowing paragraphs without internal structure
- Sentences with heavy qualifying clauses
- Content wrapped in image carousels, accordions, or video players
- Answers that require three sentences of setup before they land
The pattern is simple: AI Overviews extract content that's already extractable. That sounds tautological until you realize most pages on the internet are written like academic papers, with arguments that build over paragraphs. Those pages don't get extracted no matter how authoritative they are.
03 / Which formatting moves get you extracted?
Six formatting moves win extraction on competitive queries. Use all six.
1. The 60-word window
First 40 to 65 words: answer the implicit question of the page directly. No preamble. No "in this article we'll explore." Just the answer, in a block extractable as one chunk. The 60-word window is the first 40 to 65 words where direct answers must live. The lead block on this page is 52 words and structured to be lifted whole. It's the single highest-leverage AEO move available.
2. Question-shaped headers
Every major section heading is the question a user would type. Not "AEO Strategy" but "How do
you optimize for AEO?" Each question becomes a candidate extraction. Google's AI looks for
<h2>question?</h2><p>answer</p> patterns specifically.
Question-shaped headers are H2s phrased as the question a user would type.
They roughly double the extraction rate over topic-shaped headers on competitive queries.
3. Definitional structure
When you introduce a concept, define it in the form [term] is [definition]. This article uses the pattern when introducing position zero, the 60-word window, and AEO itself. The pattern is so common in extracted content that it's essentially required. It also doubles as the strongest LLMO move available, since large language models absorb these definitions during training cycles.
4. FAQPage schema mirroring a visible FAQ
Pages with valid FAQPage schema show up in AI Overviews at roughly 2 to 3 times the rate of pages without it on FAQ-rich queries. Critical detail: the schema must match the visible FAQ exactly. Google penalizes schema with answers that aren't visible on the page.
5. Comparison tables for comparison queries
If the post compares anything (versions, approaches, tools, frameworks), build a real
<table> with <thead> and <tbody>.
Markdown tables that render to HTML tables work too. Tables get extracted whole. Section 05
of this article uses one.
6. HowTo schema for procedural content
Less universally useful than FAQPage but extremely powerful when applicable. Sequential steps, numbered playbooks, "how to" tutorials all see large gains from HowTo markup. This article includes HowTo schema in the document head listing the six moves in this section as steps.
04 / Which schema markup actually moves the needle?
Three schema types matter for AEO. Everything else is overhead or off-topic.
Article schema is the metadata layer. Author, publisher, publish date, headline, description. It doesn't trigger extraction on its own. But Google trusts pages with proper metadata more, and trust translates to impressions in AI Overviews. Add it. Ten minutes per page. The gains compound. See Google's Article schema documentation for the required fields.
FAQPage schema is the single most powerful AEO schema. Each Q&A pair is eligible for direct extraction. The schema and the visible FAQ must match exactly. Pages with valid FAQPage schema get cited in AI Overviews at 2 to 3 times the rate of pages without it on FAQ-rich queries. If you write only one schema block, write this one.
HowTo schema is for procedural content. Sequential steps in a tutorial, a playbook, a recipe. Less universally relevant than Article or FAQPage, but where it applies, it's the highest-converting schema available. HowTo markup gets pages into the "step-by-step" rich result format, which is one of the most clickable AI Overview slots.
What's off-topic for AEO:
- Product schema: useful for e-commerce extraction, irrelevant for informational content
- Review schema: useful for product queries, doesn't trigger AI Overviews for informational queries
Add Article, FAQPage, and HowTo where they fit. That's 95 percent of the schema work that matters. Anything beyond that is yak-shaving.
05 / How should you structure H2s for AEO?
This pattern is important enough to deserve its own section.
The old way of writing section headers was topic-based. "Pricing." "Features." "Implementation." These are fine for a brochure. They're terrible for AEO. Google's AI doesn't know what question "Pricing" answers, so it doesn't know when to extract content from that section.
The new way: phrase every H2 as the exact question a user would type.
| Topic-shaped (weak) | Question-shaped (strong) |
|---|---|
| Pricing | How much does X cost? |
| Features | What does X actually do? |
| Implementation | How do you set up X? |
| Comparison | Is X better than Y? |
| Use cases | When should you use X? |
Once the H2 is a question, the paragraph immediately below it must answer that question in two to four sentences. No setup. No qualifying. Direct answer, then expansion. This is the question-shaped header pattern.
The discipline extends to every section in the article. This article uses question-shaped headers throughout. Section 01 asks why AEO matters. Section 02 asks what gets extracted. Section 03 asks which formatting moves win. Every H2 is a candidate extraction.
Get this structure right and AI Overview extraction becomes routine instead of rare.
06 / Why do some pages get extracted and others don't?
The honest pattern across the queries I track.
Pages that consistently get extracted share five traits:
- A direct answer inside the 60-word window
- At least one bulleted list per 500 words
- Question-shaped headers throughout
- Valid FAQPage schema mirroring a visible FAQ
- Total length between 1,200 and 2,800 words
Pages that consistently get ignored share these:
- Lead paragraph that sets up the topic instead of answering it
- Long prose blocks with no internal structure
- Topic-shaped H2s instead of question-shaped H2s
- Missing or broken schema
- Either under 600 words (looks thin) or over 4,000 words (looks unfocused)
The pages that ignore AEO formatting and still get extracted are almost exclusively from very high-authority domains: Wikipedia, government sites, university research pages. Everyone else has to format for extraction explicitly. There's no "I have good content and they should just find it" exception.
This is unfair if you're the writer, because it means content quality alone doesn't win. Structure does. But it's also the opportunity, because most of your competitors are still writing for 2018 SEO and their pages aren't structured for AI extraction. Format for AEO and you skip the line.
07 / What is the position-zero trade?
The honest conversation no one wants to have.
When your content gets extracted into an AI Overview, the user often gets the answer without clicking through. Your impression count goes up. Your click-through rate goes down. For a content business that measures success in sessions, this looks like a loss. This trade has a name: the position-zero trade, also called the extraction tax. You pay clicks in exchange for visibility above the traditional results.
Two ways to think about the trade.
Brand absorption play. Being the cited source in an AI Overview puts your brand in front of the user without requiring the click. Over time, this compounds into brand recognition that converts when the user has a real commercial intent later. The click you didn't get today is the conversion you do get next quarter when they've seen your name three times in AI Overviews and finally need the service.
Funnel-stage play. Top-of-funnel queries ("what is X") have low commercial intent anyway. Getting extracted on these queries costs you little because those users weren't converting through the click anyway. Save the click-through optimization for middle-funnel queries ("how do I choose X," "X vs Y for Z") where the user is closer to buying.
The brutal version: if your business model depends on top-of-funnel clicks converting directly into revenue, you have a different problem AEO can't solve. Most businesses, including mine, don't operate that way. We use top-of-funnel content for awareness and middle-funnel content for conversion. AEO optimizes the awareness layer perfectly.
The position-zero trade is a trade, not a loss. Most operators get this backwards because they're measuring the wrong thing.
08 / Frequently asked
Want a page formatted this way?
That's the service. Conversion landing pages and notes built with AEO, GEO, and LLMO from the first line. Same outcome your agency promises, in days instead of quarters. First ten clients get founder pricing.
Drop me a line →