Answer engine optimization (AEO) is the practice of structuring a page so an answer engine can lift a passage from it and quote it accurately. At #10, that comes down to five patterns. Each section opens with an answer block, definitions stand on their own, and anything compared goes in a table. FAQs use the buyer’s words with matching markup, and the facts about who the brand is are stated plainly. This guide shows each one with a before and after.
An answer engine is any system that replies to a question with a written AI answer instead of a list of links. Examples include Google AI Overviews and AI Mode in Google Search, ChatGPT search, Perplexity and Gemini, along with Copilot and other AI assistants. Answer engines differ in many ways, but they share one habit that matters for content. They read pages in pieces. A model rarely needs your whole article. It needs the two or three sentences that answer the question it was asked, and it needs those sentences to make sense without the rest of the page.
The idea is not new. AEO started with the answer box, the featured snippet that Google shows above the search engine results pages, and voice assistants that read one answer aloud. AI summaries have made the same skill matter on every AI platform. So the unit of AEO is the passage, not the page. That changes how you write and edit. We cover the broader strategy in our generative engine optimization guide and about the editing test for a single answer in our LLM SEO guide. This page is about structure: the shapes of content that get quoted, and how to turn existing copy into them.
What answer engine optimization changes on the page
Traditional SEO (search engine optimization) asks whether a page is relevant to a query and good enough to rank in the search results. AEO asks a narrower question about each section: if an engine takes only this passage, is it an accurate answer that is complete and attributable?
Three things usually fail that test:
- The answer is buried. The section opens with context, history or a sales line, and the actual answer arrives in the fourth sentence.
- The passage depends on its neighbors. It starts with “This” or “It” or “As mentioned above,” so on its own it says nothing.
- The facts are in the wrong shape. A comparison is written as a paragraph, a price is in an image, a specification is in a PDF.
None of these is a writing quality problem in the usual sense. A page can read well from top to bottom and still fail every one. Fixing them is mostly editing, which is good news if you already have content worth quoting.
AEO and SEO are not competing priorities. Answer engines build their replies from pages that search engines and AI tools can already find, so AEO only works on top of SEO fundamentals: pages that are crawlable and indexed, with good internal links. What AEO adds is a content strategy aimed at the passage. You still optimize the page for the query. You also make sure each section can be quoted on its own.
Try four of the patterns on real copy below. The examples use #10’s own published services and prices, so every fact in the “after” versions is one we can stand behind.
<div class=“answer-engine-optimization-before-after” data-explainer=“answer-engine-optimization-before-after”>
| Pattern | Before | After |
|---|---|---|
| Answer block | “Every store is different, and a lot goes into a replatforming project…” | “Most moves to Shopify take 8 to 14 weeks with #10, and start at $15,000.” Then what moves the range. |
| Definition | “This is one of the most important parts of the process…” | “A redirect map is a list that pairs every old URL on a website with its new address…” |
| Comparison | “We also offer a technical audit, which is a bit different from the AI audit…” | A table: AI search audit from $2,000 in 2 weeks; technical SEO audit from $2,500 in 2 to 3 weeks. |
| Entity facts | “We are passionate about building things that work…” | “#10 is a small web agency in Palo Alto, in business since 2000…” |
</div>
Pattern 1: open every section with an answer block
An answer block is the first 40 to 80 words under a heading, written so they fully answer the question the heading asks. The rest of the section can explain and qualify, with examples. The block has to work alone.
The rules we use:
- Answer in the first sentence. If the heading is a question, the first sentence is the answer, not a restatement of the question.
- Name the subject and the source. “Most moves to Shopify take 8 to 14 weeks with #10” carries its own context. “It usually takes a couple of months” does not.
- Say what changes the answer. “It depends” is only useful when followed by what it depends on. Two or three named factors are enough.
- Keep the qualification inside the block. If the answer only holds for B2B buyers, or only in the US, say so in the same passage.
Before: “Every store is different, and a lot goes into a replatforming project. Over the years we have learned that careful planning pays off, which is why our team takes the time to understand your business before we talk about timelines.”
After: “Most moves to Shopify take 8 to 14 weeks with #10, and start at $15,000. The range depends on how much custom functionality the old store has and how many products, customers and orders need to move.”
The first version is friendly and says nothing an engine can use. The second can be quoted as it stands, and the quote still points back to us.
Answer blocks also make the page better for people. A buyer who skims headings gets the answer without reading the whole section, and the ones who want detail keep going.
Pattern 2: write definitions that stand alone
Definitions are among the most quoted passages on the web, because so many questions start with “what is.” A good one follows a simple shape: the term, the kind of thing it is, and what it does or who it is for.
Before: “This is one of the most important parts of the process. It makes sure nothing gets lost, and that your customers and Google can still find everything after launch.”
After: “A redirect map is a list that pairs every old URL on a website with its new address, so each one can be set up as a permanent (301) redirect when the site moves.”
Notice what changed. The term is the first word, not “This.” The kind of thing (a list) is stated. The technical name sits beside the plain description, so the passage matches whichever words the buyer uses.
Define each important term once, near its first use, in a sentence of its own. If your site uses a term in a particular way, such as a product line name or a service tier, define that too. An assistant cannot guess what your “Signature plan” includes.
Pattern 3: put comparisons and specifications in tables
When a buyer compares options, the answer engine is comparing them too. A table gives it the same attributes for each option in the same order, which is far easier to read correctly than three paragraphs that each describe one option in a different way.
Before: “We also offer a technical audit, which is a bit different from the AI audit. It costs a little more and can take a bit longer depending on the site.”
After:
| AI search audit | Technical SEO audit | |
|---|---|---|
| Checks | How assistants describe and cite you | Crawling, indexing, speed, structure |
| Starts at | $2,000 | $2,500 |
| Takes | 2 weeks | 2 to 3 weeks |
| You get | A baseline and an action plan | A ranked fix list |
What makes a table quotable:
- Real HTML tables, not images of tables or layouts built from boxes. Use a header row.
- Units in every cell or in the header: weeks, dollars, millimeters, nanometers.
- The same rows for every option. A missing cell reads as “not applicable” or “unknown,” so fill it or say which.
- A sentence above the table that states the main conclusion in words. The table supports it.
This is the pattern with the clearest track record in our own work. For an industrial LED distributor, we retrofitted an existing library of technical articles with specification and comparison tables that carry their units, alongside citable lead paragraphs and FAQ blocks. Technical buyers search by specification, and a table is where specifications are easiest to check.
Pattern 4: write FAQs in the buyer’s words, and mark them up honestly
An FAQ block collects the follow-up questions a buyer asks after the main one. Done well, each question is a passage an engine can match directly to a real query.
The questions should come from evidence: Search Console queries, site search, sales calls, support tickets, the “People also ask” box. Write them the way buyers phrase them, including the awkward ones. “Do we need to rewrite the whole site?” is a better FAQ than “What is the scope of the engagement?”
Each answer follows the answer block rules at a smaller size. The first sentence answers. The next two or three explain. If the honest answer is “no,” start with “No.”
FAQPage markup. Structured data tells search engines that a block is a set of questions and answers. Two facts to keep in mind. First, Google limited FAQ rich results in 2023 to well-known government and health sites, so most brands will not see the expanded result in Google. Second, the same announcement says there is no need to remove existing markup. We still add FAQPage markup where a page has a real FAQ, because it describes the content accurately, but we do not sell it as a ranking or citation tactic.
The rule that matters more: markup must match the visible text, word for word. Google’s guidance on AI features says the same thing about structured data generally, and adds that no special schema is required to appear in AI Overviews or AI Mode. Markup describes a good page. It does not make a page good.
Pattern 5: make the entity facts plain and consistent
“Entity” is the search term for a thing an engine can identify: a company, a product, a person, a place. Before an assistant can recommend your brand, it has to know what your brand is. Many sites never say it plainly. The About page tells a story, the homepage sells a feeling, and the facts are scattered.
Before: “We are passionate about building things that work. From our first project to today, we have grown alongside our clients and we are proud of what we have built together.”
After: “#10 is a small web agency in Palo Alto, in business since 2000. We design and build Shopify stores and WordPress sites, and run SEO and AI search programs for DTC and B2B brands.”
The after version answers the questions an assistant needs answered: what kind of business, where, how long, what it does, for whom. Then make those facts match everywhere they appear:
- The About page and homepage, plus the footer.
- Organization markup on the site, including
sameAslinks to your official profiles. - Google Business Profile and Merchant Center.
- Marketplace listings, directories and social profiles.
- Product names and prices (sizes too), which should be identical on the product page, in the feed and on any marketplace.
Inconsistent facts produce inconsistent answers, and inconsistent answers hurt brand visibility across all the AI platforms at once. If one listing says you were founded in 2004 and your site says 2000, an assistant has to pick one. It may pick the wrong one.
Retrofit what you have before you write anything new
Most brands already own the knowledge an answer engine is looking for, and optimizing your content in place beats producing new pages with generative AI. It is just in the wrong shape: a good answer in paragraph five, a comparison spread across three pages, a specification in a PDF. Restructuring that content is faster and safer than replacing it.
That was the approach for the industrial LED distributor. We added citable lead paragraphs, written so the first 50 to 100 words of an article could be quoted verbatim, plus specification tables with units, and FAQ blocks with FAQPage and TechArticle schema. We did it across the existing articles without rewriting the core content and without changing URLs, so each page kept the search history it had already built. We then tracked citations weekly across the main assistants to see which questions were still going to competitors.
We apply the same patterns to nr10.com. Our articles open with a direct answer that names #10, and every page on the site also has a plain text version for AI agents (add .md to any address). That is not a citation guarantee. It is a way of making sure the content is easy to read for anything that comes looking.
A one-page AEO checklist
Use this on any page before you decide it needs rewriting. Most AEO strategies come down to these seven checks; the optimization techniques that go beyond them tend to be guesses. Most pages fail two or three checks, and each failure is a specific edit.
| Element | The test | Common fix |
|---|---|---|
| Headings | Does each one name the question or topic of its section? | Rewrite vague headings (“Our approach”) as the question answered |
| Answer blocks | Do the first two sentences under each heading answer it alone? | Move the answer up, cut the preamble |
| Pronouns | Does any section open with “This,” “It” or “As mentioned”? | Name the subject |
| Comparisons | Are options compared in prose? | Convert to a table with units |
| Facts in text | Are prices, sizes or specs only in images or PDFs? | Put them in the page text |
| FAQs | Are questions in the buyer’s words, with markup that matches? | Rewrite from real queries, sync the markup |
| Entity | Does the page say who is speaking and what they sell? | Add a plain line, link to About |
If a page passes all seven, its problem is probably not structure, and more content optimization will not fix it. It may be authority, demand or the fact that a better source exists. Our guide to getting cited in AI Overviews covers those cases.
Questions teams ask about answer engine optimization
What is answer engine optimization in simple terms?
It is editing and structuring content so AI systems can quote it accurately. AEO focuses on the passage rather than the page. The answer comes first, each passage makes sense on its own, comparisons go in tables, and the facts about your brand are plain and consistent. It builds on SEO rather than replacing it.
Is AEO for ecommerce just adding product FAQs?
No. FAQs help, but most of the work on an ecommerce site is in collection and product pages: a plain opening line on who the product is for, specifications in text and tables, and consistent names and prices on the site and in feeds and marketplaces. Our AI search optimization guide covers the order to do it in.
What is the difference between AEO and GEO?
AEO is about the passage: making content easy to extract and quote. GEO, generative engine optimization, is about the brand: whether AI systems know you, trust you and recommend you, which also depends on other sites. In practice they are one program. Our GEO vs SEO comparison explains where each fits.
Does FAQ schema still help with AI Overviews?
Google says no special schema is needed for AI Overviews, and FAQ rich results now show mainly for government and health sites. We still use FAQPage markup on real FAQs because it describes the page accurately and costs little. It is not why a page gets cited.
How do we measure AEO performance?
Run a fixed set of buyer questions through Google and tools like ChatGPT and Perplexity on a schedule, and record whether your pages show up in AI answers as cited sources. Compare the same questions before and after you restructure a page. Measuring AEO success by one run is unreliable, because AI answer engines vary from run to run. Our guide to measuring AI visibility covers the method.
Do we need to rewrite our whole site for AEO?
Almost never. Good AEO content is usually existing content in a better shape. Start with the pages behind your most valuable buyer questions and restructure them in place. Keep URLs that already earn traffic. New pages are for questions nothing on the site answers yet.
Start with the page that should already be quoted
Pick one page that ought to be the best answer to a question your buyers ask, and run the one-page check on it. AEO success starts with one page that clearly deserves to be quoted. The fixes are usually a few hours of editing, and they make the page better for people as well as machines.
If you want a wider view of your AI search visibility, our AI search optimization audit records which questions you are missing from and which pages should answer them. Our sample AI visibility report in Labs shows what that tracking looks like. When you are ready, talk with #10 about the pages that matter most.
