Answer Engine Optimization (AEO) Fundamentals

Track: AEO & GEO · Format: text lesson · Est. read: 15 minutes

By the end of this lesson you'll be able to take one real buyer question and structure a single page so an answer engine can find the answer on it, lift it cleanly, and cite you as the source.

This is the second course in the AEO & GEO track. It assumes you've done AI Visibility 101, so you already know that answer engines read a few pages and write back a conclusion, and that a brand shows up only when the engine can find, read, and trust its page. That was the mental model. This is the craft: the page-level work that turns "findable, readable, credible" into a page that actually gets quoted. The discipline has a name, answer engine optimization, and everything below is a piece of it.

We'll build the whole thing around a single question a real buyer might type, and by the end you'll have watched a page get assembled from scratch to answer it.

Pick the question, not the keyword

Start here, because everything downstream depends on it. Old SEO started from a keyword: a short phrase like "sales analytics pricing" that you tried to rank for. AEO starts from a question, the actual sentence a buyer says to ChatGPT.

Those are different inputs. A keyword is two or three words stripped of context. A question carries the whole situation: who's asking, what they already know, what they're really trying to decide. "Sales analytics pricing" tells you almost nothing. "How much does a sales analytics tool cost for a 20-person team, and is it worth it over a spreadsheet" tells you the team size, the budget anxiety, and the real alternative they're weighing. The engine gets the second version, so you should optimize for the second version.

Here's the practical move. Before you touch the page, write down the exact question in the buyer's own words. Not your marketing paraphrase. The messy, specific sentence they'd actually type. If you don't know what that sentence is, that's the first thing to go find out: ask your sales team what buyers actually ask on calls, read your support tickets, or type a few guesses into ChatGPT yourself and watch what it does with them.

One page should own one question. A page trying to answer six questions answers none of them cleanly, and the engine, hunting for a passage that resolves this question, gives up on the page that hedges across all six. Which question a page owns, and whether you own enough of your buyers' questions across your whole site, is prompt coverage, and it's the map you're filling in one page at a time.

For the rest of this lesson, our question is:

"What's the best way to track B2B pipeline for a small sales team that's outgrowing spreadsheets?"

Hold that sentence. We're going to build a page that answers it.

Structure the page to answer it directly

Now the habit that matters most in AEO. Lead with the answer.

A retrieval system pulls a passage that stands on its own and resolves the question. If your answer arrives in paragraph five, after the origin story and the mission statement, the engine often can't isolate it, and it quotes a competitor who put the answer in the first line. Humans skim past a slow intro. A machine trying to lift a clean, quotable sentence just leaves.

So the first sentence under your heading should answer the question well enough to be quoted with nothing else around it. Everything the buyer needs, in one self-contained line, before any context.

Watch the difference. Same page, same facts, two openings.

Weak opening:

At [Company], we've spent years thinking deeply about the challenges growing sales teams face. Our journey began when our founders, frustrated with the tools available, set out to build something better. Pipeline tracking is a critical part of any modern sales motion, and there are many factors to consider when choosing an approach that's right for your unique needs.

Read that as an engine would. Nothing in it answers the question. There's no method, no recommendation, no fact to lift. It's four sentences of throat-clearing, and the retrieval step gets nothing to quote.

Strong opening:

The best way to track B2B pipeline for a team outgrowing spreadsheets is a lightweight CRM with a shared pipeline view, so every rep sees deal stage and next step in one place instead of in scattered tabs. Spreadsheets break down around 5 to 10 active reps: version conflicts, no reminders, and no single source of truth. A dedicated pipeline tool fixes those three specifically.

That opening answers the question in the first sentence, names the failure mode in the second, and sets up the rest of the page in the third. An engine can quote sentence one and be done. A buyer skimming can stop after it and know what you think.

The shape to aim for: heading that mirrors the question, direct answer in the first line, then the supporting detail. Use the buyer's own phrasing in the heading (## How to track B2B pipeline when spreadsheets stop working) so the engine's question and your heading line up. Short definitional passages and the occasional table beat an answer buried mid-paragraph, because they're easy to isolate. If part of your answer is a set of steps or a comparison, format it as steps or a table, not as prose the engine has to reverse-engineer.

This is 80% of the work. The rest sharpens it.

Add schema, but only the schema that fits

Structured data is code you add to a page (in the schema.org vocabulary, usually as JSON-LD in the head) that states what the content is. FAQPage says "these are question-and-answer pairs." HowTo says "these are steps, in order." Organization says "this is the company, here's its name and description." You label the meaning instead of leaving the engine to infer it from your formatting.

It's one of the few AEO moves that sits fully in engineering's hands and ships this week. The catch is that most teams cargo-cult it: they paste every schema type they can find onto every page, which does nothing and can actively hurt when the markup describes content that isn't there. Match the type to what the page actually is.

The three that earn their place for most B2B content:

• FAQPage when the page genuinely holds a list of questions and answers. Mark up a real pricing FAQ and you hand the engine a tidy set of Q&A pairs to lift. Do not wrap a single blog post in FAQPage because you invented three fake questions to qualify. Engines and Google's guidelines both frown on that, and it can get the markup ignored.
• HowTo for a genuine step-by-step process, like "how to set up pipeline stages." It tells the engine these are ordered steps, so it can pull the sequence cleanly. Our pipeline page has a real setup sequence, so this one fits.
• Organization on your homepage or about page, once, with your name, logo, and description. This is the one that feeds the knowledge graph, the entity-level picture of who you are, which helps engines connect a citation back to your brand.

Keep the markup honest and in step with what a visitor sees, and run it through a validator (Google's Rich Results Test, or the schema.org validator) before it ships. Schema won't rescue a page with no real answer on it. It makes a good answer easier to pull, quote, and attribute. Ship the answer first, then label it.

Make the engine trust the page

You can write the cleanest answer on the internet and still lose the citation to a page the engine trusts more. When two pages say roughly the same thing, source authority breaks the tie: the signals that tell the engine this page is worth standing behind.

For a single page, that comes down to a few concrete things you can add today:

• Put a real, named author on it, with a one-line bio that says why they'd know. "By Dana Ruiz, who ran sales ops at two Series B startups" beats an anonymous post making the identical claim. This is the E-E-A-T side of authority, the human-credibility half.
• Cite your numbers. If you say spreadsheets break down at 5 to 10 reps, say where that figure comes from, your own data or a named source. A claim with a citation outranks the same claim floating unsupported.
• State honest tradeoffs. A comparison page that admits where a competitor is genuinely better reads as more trustworthy than one that pretends you win on everything, and engines increasingly reward that balance.
• Be consistent about who you are across your site and the wider web, so the entity picture stays clean. Same name, same description, same facts everywhere.

Authority compounds slowly, over months of consistent, cited, credited work. But the page-level pieces (a named author, a cited number, an honest comparison) are things you add before you hit publish, and they're often what decides whether your clean answer or someone else's gets the quote.

Signal crawler access with llms.txt

One more lever, newer and lower-stakes: llms.txt.

It's a plain Markdown file at the root of your site (yoursite.com/llms.txt) that hands AI crawlers a short, curated map of your best pages. Where robots.txt tells a crawler what it may not touch, llms.txt tells an AI system what deserves its attention: here's my pricing, here are my docs, here's the comparison page, each with a one-line description and a link to a clean version. You're shaping the model's first impression instead of trusting it to rebuild your priorities from a sprawling site.

Be honest about where this stands. Support across engines is partial and still shifting, so treat it as a cheap bet, not a settled standard. It costs an afternoon to write and it runs with the grain of where AI discovery is heading. List your highest-value pages with a line of description each, link to versions that are easy to retrieve and read, and keep it current. A curation file full of dead links defeats its own purpose. If you have to choose where your hours go this week, the direct-answer rewrite from earlier beats this. Get to llms.txt once the pages themselves are right.

Worked example: building the page live

Let's assemble the whole page for our question, top to bottom, so you can see every lever land in one place.

The question: "What's the best way to track B2B pipeline for a small sales team that's outgrowing spreadsheets?"

Step 1: name the reader and the real alternative. A sales lead or founder, 5 to 15 reps, currently on spreadsheets, feeling the pain but not sure the switch is worth it. Their real alternative isn't a rival CRM. It's staying on spreadsheets. So the page has to answer "is this worth changing at all," not just "which tool."

Step 2: write the heading to mirror the question.

## How to track B2B pipeline when spreadsheets stop working

Step 3: lead with the self-contained answer. First line resolves the question on its own:

The best way to track B2B pipeline for a team outgrowing spreadsheets is a lightweight CRM with a shared pipeline view, so every rep sees deal stage and next step in one place. Spreadsheets usually break around 5 to 10 active reps, where version conflicts and missed follow-ups start costing real deals.

Step 4: lay out the body so each part is liftable.

• A short paragraph on when spreadsheets actually stop working, with the specific signs (two reps editing the same file, no reminders, no reporting). This owns the "is it worth it" half of the question.
• A small table comparing spreadsheet vs. lightweight CRM vs. full enterprise CRM across three columns a buyer cares about: setup time, cost per rep, when it fits. A table because the engine can lift a row cleanly.
• A genuine step sequence: "Setting up your first pipeline in an afternoon," five ordered steps. This part gets HowTo schema because it's a real ordered process.

Step 5: add the authority signals. A named author with sales-ops experience and a one-line bio. The "5 to 10 reps" figure attributed to a source (your own onboarding data, say). An honest line admitting that a team under 5 reps might genuinely be fine on a spreadsheet for now. That honesty makes the rest more believable.

Step 6: label it with schema. HowTo on the setup sequence. FAQPage only if you add a real cluster of buyer questions at the end ("Do I need a full CRM?", "Can I migrate my spreadsheet?"), each answered directly. Organization already lives on your homepage, so you don't repeat it here.

Step 7: add the page to llms.txt with a one-line description once the page is live and right.

Read the finished page back as the engine sees it. The question maps to the heading. The first line is a quotable answer. The comparison is a liftable table. The steps are labeled as steps. A credible author stands behind a cited number. That page doesn't need to outspend anyone. It needs to be the passage the engine finds easiest to quote and easiest to trust, which is the whole job.

The one idea to take away

A page gets cited when it leads with a self-contained answer to one specific buyer question, formats the supporting detail so an engine can lift it, and carries enough authority (a named author, a cited number, an honest tradeoff) that the engine trusts it over the alternatives. Schema and llms.txt sharpen a page that already does that. Neither rescues one that doesn't.

Everything else in AEO is a variation on those moves, applied one buyer question at a time until you own the answers across your whole site.

What's next

You now have the page-level craft. The track continues:

• Reading an AI Visibility Report turns this into measurement: how to tell whether your work is landing, framed around citation share.
• GEO vs. SEO gives marketers who already know search the precise delta between the two.

Together these build toward the AI Visibility Analyst certification, which tests whether you can look at a page, spot why it isn't getting cited, and fix it.

Before you move on, take the quiz in quiz.md. It's applied, and the short-answer rewrite is the fastest way to find out whether the "lead with the answer" habit actually stuck.