The Questions Your Buyers Are Asking AI

AI prompts don't look like search queries

Watch someone use Google and then watch the same person use ChatGPT. The behavior is completely different.

In a search box they type three or four words, because they've learned that's what works. In a chat window they write a sentence or two, in their own voice, including the context and the worry. They ask the question they'd ask a friend who happened to know the industry.

Those right-hand questions are the ones your buyers actually have. They just never asked Google, because Google was never going to answer them properly.

This is Signal 5 of the 8 Signals, worth 10 of 100 points. It's not the biggest category, but it may be the one with the shortest distance between doing the work and seeing the result — because these are questions you already answer, out loud, every week.

What answer engine optimization actually is

Answer engine optimization, or AEO, is the practice of structuring a website so that AI assistants and answer engines can find and use direct answers to the questions buyers actually ask, rather than only ranking the page in a list of links. Where traditional SEO competes for position, AEO competes for the answer itself.

It overlaps heavily with what people call GEO, or generative engine optimization, and with plain good content strategy. The distinction that matters in practice: AEO is specifically about question-and-answer coverage — whether a direct answer to a real question exists on your site, in a form an engine can lift.

The five topics buyers care about most

You don't have to invent this list. Marcus Sheridan mapped it years ago as the Big 5 — the five subjects buyers research hardest before they'll talk to anyone.

• Cost and price. What does this cost, and what makes it cost more or less?
• Problems. What goes wrong with this, and what are the downsides nobody mentions?
• Versus and comparisons. This option or that one? Your approach or the other approach?
• Reviews. What is this product or provider actually like to work with?
• Best in class. Who or what is the best at this, for someone in my situation?

That framework predates AI search by more than a decade, and it turns out to describe AI prompts almost exactly. It's the same insight from a different direction: buyers have always wanted these five answers, and they've finally got a tool that will give them one.

Notice how many of the five most businesses actively avoid. Price, because it depends. Problems, because why advertise your flaws. Comparisons, because that means naming competitors. Best in class, because you can't very well say it's you.

The price question, and why avoiding it costs you

Price is the single most-asked question in every industry I've worked in, and the one most sites refuse to touch. The reasoning is always the same: it depends on the job, and a number without context will scare people off or get used against us.

Both concerns are real. Neither is solved by silence.

What silence actually does is send the buyer somewhere else for the answer — and now a competitor, or a generic industry article, is framing your pricing for you. Worse, in an AI answer, the business that explains its pricing clearly looks like the transparent one, and you look like the one that wouldn't say.

You don't have to publish a rate card. You have to answer the question in the shape it was asked. A range plus the five factors that move it within that range is a complete, honest answer that costs you nothing and disqualifies the people who were never going to buy anyway.

Mining your sales calls for the real list

The best question list isn't from keyword research. It's from the last twenty conversations you had with prospects.

Here's the exercise. For a couple of weeks, write down every question a prospect asks you — including the ones you're tired of answering, especially the ones that make you wince. Then sort them by how often they came up. That ranked list is your content plan for this signal, and it's better than anything a tool would have given you, because it's the actual language of the actual buyer.

Two patterns worth watching for. The question that gets asked on nearly every call is usually the one that deserves its own page rather than an FAQ line. And the question you find yourself dreading is almost always the highest-value one to answer publicly, because your competitors are dreading it too.

This is the same extraction problem behind the proof signal — the material exists as conversation and has never been written down.

How to write an answer that gets used

Three rules, and the first one carries most of the weight.

Answer in the first sentence. Not context, not "great question," not "it depends." The direct answer, immediately, then everything that qualifies it. A hedge at the top reads to an engine as no answer at all, and it reads to a human as evasion.

Use the question as the heading, in the buyer's words. "How much does a roof replacement cost in Cincinnati?" not "Pricing." If people phrase it several ways, pick the most common one and cover the variants in the body. More on the shape of this in how to write content AI can quote.

Make each answer stand alone. Someone arriving at that one answer, with no other context, should get a complete response. That's how an engine will encounter it too.

A real FAQ built this way does double duty: it serves the person who scrolled to it, and it gives an engine a set of clean question-and-answer pairs it can lift with confidence.

Marking it up so it can be found

Once the answers exist, FAQPage schema tells a machine explicitly that this block is a question and that block is its answer — rather than leaving it to infer the relationship from the layout.

One rule governs it: the answers in the markup must match the answers on the page word for word. Schema containing an answer a human reader can't find is a contradiction, and contradictions cost you on entity clarity. The mechanics of adding it are covered in Signal 6.

What I don't know yet

I don't know how AI assistants weigh a question answered on your site against the same question answered on a large publisher's site. I'd expect the publisher to win on general questions and you to win on ones specific to your market, but I can't prove it.

I don't know whether FAQPage schema still carries the weight it did, given how much Google has changed about how it displays FAQ results over the past few years. I keep using it because it's cheap and it clarifies structure, not because I can measure a lift.

What's held: the businesses that answer real questions plainly get named in answers about those questions. I'll update this as more becomes measurable.

Frequently asked questions

Related reading

• The 8 Signals — the full framework, all eight scored categories, and what your total means.
• How to Write Content AI Can Quote — Signal 4, and the shape an answer has to take to get used.
• Why Thin Coverage Keeps You Out of AI Answers — Signal 3. The biggest questions deserve pages, not FAQ lines.
• The Proof AI Is Looking For — Signal 1, and the same extraction problem in a different form.
• llms.txt, Schema & the Files AI Looks For — Signal 6, including how to add FAQPage schema.

This is Signal 5 of 8. See the full framework →