AEO Lessons That Changed How We Work

The first two pieces in this series covered the why and what. This is a look at the FAQ-page approach we used, and how that translates to senior living and home care.

If you've been paying attention to how AI-powered search works, you've heard that content needs to be "optimized for AI." But what does that actually look like in practice? For us, it came down to the FAQ page more than anything else, and it took our AEO visibility from 19.62% to 64.45%.

This isn't about FAQs being trendy. It's about building pages that take advantage of how answer engines actually read content. The same logic applies whether you're a SaaS company like WelcomeHome building 33 FAQ pages or a senior living community answering the questions families ask when they're scared and short on time. Here's how we built ours.

How AI engines read content

Before getting into the how, the why matters enormously.

When a human reads a web page, they skim, scroll, and pick up context from earlier sections. An AI answer engine works differently. It processes content at the passage level, scanning for specific, self-contained blocks of text it can extract and present in a response. It's not reading your page top to bottom. It's hunting for the clearest, most citable piece of information it can find.

On top of that, there's a behavior called query fan-out. When a user asks an AI tool a question, the AI breaks it into several smaller, more specific sub-questions, searches for the best answer to each, and stitches everything together into a single response.

What that means for content: a single user question might trigger five or six sub-queries simultaneously, each from a different source. That's both a challenge and an opportunity. 

Why FAQ pages are built for AEO

FAQ pages are the most AI-extractable content format available, not because they look clean, but because of FAQ structure. 

• Each Q&A is a citation target. Since each answer is self-contained and matches a specific sub-question, it can be lifted and cited on its own. Since AI tools use query fan-out to break a single user question into several smaller ones, an FAQ page can capture multiple citation opportunities from one page. 

• The format is self-contained. AI systems read content in passages rather than full pages. FAQ format naturally forces self-contained answers. Each entry has to make sense on its own, without requiring the reader or AI to have read what came before it.

• Question-based headers are direct query matches. When your H2 is phrased as a question and a buyer types that same question into ChatGPT or Perplexity, your page sends a direct relevance signal. Vague headers like "Overview" or "Features" force the AI to guess whether your content is relevant. 

• FAQ Page schema amplifies visibility. Adding FAQPage structured data markup signals to AI systems exactly what kind of content they're looking at, reducing guesswork and increasing citation likelihood.

How do you find out what questions buyers are asking?

The foundation of a strong FAQ page is knowing what questions people are actually asking. We pulled from every available source: company documents, one-pagers, knowledge bases, decks, website content, and CRM data. We also looked at how competitors were using FAQs across their sites. Competitive analysis revealed that the strongest players were building pages that mirrored the exact language buyers use when searching in AI tools.

Topics were organized across the full buyer journey: general product questions, pricing, integrations, onboarding, ROI, comparisons, geography, and more. 18 topics for one product line and 15 for another, each mapped to where a buyer might be in their decision process.

How do you draft accurate FAQ content with AI?

We used AI tools to generate initial FAQ drafts by pulling directly from source documents. This wasn't about letting AI write blindly. It was about using AI as a research and synthesis layer to surface accurate answers quickly, then expanding and refining from there. Each draft was built around AEO structural principles: answer-first, self-contained, specific, and connected back to the product. 

How should FAQ pages be structured for extractability?

We built the actual landing pages in HubSpot, structuring headers and body content to reinforce extractability. Every H2 and H3 was specific enough that an answer engine scanning the page immediately knows what that section answers. 

Structure alone isn't enough to know if it's working, though. We created a library of 50 priority search prompts and monitored them through HubSpot's AEO Dashboard. We tracked which prompts we showed up for, which competitors we were losing to, and where a page needed a rewrite rather than a fresh one. Extractability is just a starting point; the dashboard told us whether the structure was actually earning citations.

What makes a meta description work for AEO?

Meta descriptions are often an afterthought, but in AEO, they’re another extraction point. Each meta description should be written as a genuine summary of what the page delivers; 155 characters, specifically describing the content rather than a marketing tagline.

Can AI-generated FAQ drafts be trusted for accuracy?

Only as a starting point. AI-assisted drafts should be built from your own source documents, e.g., decks, knowledge bases, CRM data, and reviewed by someone with direct expertise before publishing, since accuracy and named expertise are exactly what AI engines evaluate when deciding whether to cite you.

Why does URL structure matter for AI citations?

Every FAQ page was published with a clean, descriptive URL that signals the page's topic before anyone clicks. Combined with a clear title tag and specific headers, a clean URL reinforces the other relevance signals AI uses to evaluate a page. 

What pages perform the best?

A few principles show up consistently in the pages that perform best: 

• Write for systems to enter at any section. Every H2 needs to make sense without the paragraphs before it. Avoid transitions like "as mentioned above." If AI pulls that passage out of context, it breaks. 

• Each FAQ entry should address a distinct sub-question. Don't restate the same point differently. Each entry should answer one distinct sub-question in its opening sentence, then add context. Length should match the complexity of the question.

• Answer first, context second. It’s better to lead with the direct answer in the first sentence. AI systems are more likely to cite passages that open with a clear, complete answer.

• Connect insights to the product. A page that delivers educational value but never ties those insights to what you do is training AI on a topic without giving it a reason to recommend you. Every major section should make the brand connection clear.

What AEO changes for you

The shift from "does our content rank?" to "can AI extract and cite this passage?" changes how you write, structure pages, and measure success. FAQ pages are the format that most directly aligns with how answer engines process content. 

This took time, iteration, and workflow, and the result speaks for itself: we more than tripled our AI visibility score in a matter of months. The logic behind every decision traces back to the same place: write content that AI can find, extract, and cite, and make sure that when it does, your brand is the answer.

The same approach applies directly to a senior living community or home care agency's website. Instead of pricing, integrations, and onboarding, your topics should answer the questions families ask when they're scared and short on time: what's the difference between assisted living and memory care, what does a home care aide do during a visit, how much does this cost, is there availability now. Build a real FAQ page around those questions, sourced from your own admissions team, care staff, and the calls your front desk fields every week, then structure each answer to stand on its own, and you've built the single most citable page on your site.

It's already starting to happen for operators who are ahead of this. During a Q&A at Living Room Live 2026, one attendee asked ChatGPT what the best memory care community in her city was…and it named her own community back.