What are the best AEO tactics for the B2B awareness stage?

HubSpot's outline of the buying stages defines awareness as the point where "the buyer realizes they have a problem or need." The buyer is feeling a symptom and looking for information that helps them understand it and give it a name. By the consideration stage they've named the problem and are researching ways to solve it. At the decision stage they're choosing a provider.

A buyer doesn't always move through those stages in order. The stages are still useful for sorting questions, because each kind of question needs a different kind of answer.

Suppose a company sells customer-support software. The same subject produces three different questions, depending on where the buyer is.

The awareness question in that table doesn't name a product category. We'd look for questions like it in the places where buyers describe the problem themselves, such as first sales calls and support conversations.

Lean Labs builds its AEO work from that kind of material. Its service page says the question set comes from sales calls, CRM insights and Search Console demand. Its beginner roadmap for AI search starts with one question customers ask, written the way they'd ask it. For the support-software company, that means writing down the sentence a buyer used about the backlog and making it the question the page answers.

Google's guide to generative AI features in Search says unique, first-hand content is likely to influence a site's presence in those features more than anything else in the guide. It contrasts that with content built on common knowledge.

Suppose the support-software team writes the backlog page. A common-knowledge version lists tips for working through tickets faster. A first-hand version opens with a direct answer: with volume flat, a backlog grows when fewer tickets are closed than arrive. That can come from tickets taking longer to resolve, or from less capacity to work them.

The page then gives the reasons the team has seen in its own customers' accounts, such as more reopened tickets, more handoffs between teams or unfilled shifts. It also says which report would separate one from another. A page like that gives the reader possible causes and a way to narrow them down, and it draws on evidence a summary of other articles wouldn't have.

Put the answer in the opening sentence and the explanation under it, so a reader who stops early still has something to use.

We'd keep the problem page mostly on the problem. A product mention can belong there when it helps explain a cause or a fix. The risk is a page that turns into a product description and leaves the question half answered.

A link to the next question does more for the reader. In the example, that's whether to fix the backlog with self-service content or with more staff, and the product category comes up there on its own terms. Lean Labs' roadmap grows a site the same way. Once the first page works, it adds a small cluster of pages built from the follow-up questions the first one leads to.

There's a limit on how far to take this. Google's guide says its systems understand synonyms. It also treats pages made for every variation, mainly to sway rankings or AI responses, as scaled content abuse. We'd write one page for each distinct problem. Five rewordings of the backlog question would add pages without adding information.

A buyer's problem also gets explained in places you don't own. Google's guide says its AI features can draw on what other sites, videos and forums say about a product, and it advises against chasing mentions that aren't authentic. So take the same explanation to a place where the problem is already being discussed, such as a trade publication or a community thread, and post it under your own name. We go further into this in what off-site signals matter for AEO.

Search Console's generative AI performance report gives impressions in AI Overviews and AI Mode by page, which shows which problem pages are appearing there. Analytics shows the assistant visits it recorded on those pages, and whether those visitors went on to another page. Your CRM shows contacts and deals. We set out the reports in how to track traffic and referrals from AI answer engines.

An impression on a problem page isn't a demo request, so we'd report the three as separate counts and read them together. We cover how the questions change later on in how far AEO extends into the buying process. Read together, the counts keep a problem page that's being shown and read from being written off because few deals are attached to it yet, and a page with none of the three stands out.