How to structure a B2B site so AI quotes it. The technical checklist.

Where do I start if I want AI to quote my site?

Start with your blog, because that's where most of the opportunity sits, then move to your main pages. Your blog is where the question-and-answer content lives, and question-and-answer content is exactly what answer engines lift. Your homepage and product pages matter for identity and trust, but they're rarely the thing quoted in a reply to "how do I solve X". So fix the blog first, get the compounding going, then tidy the rest.

The whole game is legibility. You're not tricking anything. You're making genuinely good answers easy for a machine to find, parse, trust and lift. Everything below is in service of that one idea.

What's the technical checklist for each blog post?

Every post needs the same handful of things, and once it's a template it costs you almost nothing per post. Here's the checklist I run:

• JSON-LD schema on every post. This is the structured data that tells a machine what it's looking at: who wrote it (author), what's a quote (marked as a quotation), and what your FAQ answers are (each as a sub-question with its answer). Without it the engine is guessing; with it, you're handing it labelled pieces.
• An answer-first TL;DR at the top. Two to four lines that answer the post's core question outright, before any scene-setting. A skim-reader and a model can both extract it without reading on, which is the point.
• Question-shaped H2s with the answer directly beneath. Every heading is a real question a buyer would type, and the first sentence under it answers that question. Detail comes second. This is the single highest-leverage habit on the list.
• Real FAQs. Three to five genuine follow-on questions with tight answers near the end. They feed both featured snippets and generative replies, and they're the easiest thing for an engine to quote cleanly.
• Attributed quotes. A named source saying something specific, marked up as a quotation. The Princeton evidence is that quotes and citations lift AI-answer visibility, so this isn't decoration; it earns its place.
• Clean, fast, crawlable pages. The content rendered in the HTML (not locked behind a script the crawler skips), a page that loads quickly, nothing blocking the crawler. If a machine can't reach it or render it, none of the above counts.

You're not tricking a machine into quoting you. You're making a genuinely good answer impossible to misread.

What about my main pages?

Give them basic org and product schema, and leave it there. Your homepage, about page and product pages need structured data that states plainly who you are (organisation schema: name, logo, what you do) and what you sell (product or service schema). That's what lets an engine connect a quote from your blog to a real, identifiable company it can name and trust. You don't need the full FAQ-and-quote treatment on every page; you need clean identity signals.

The mistake I see is people pouring effort into elaborate schema everywhere and none into the content structure that actually gets quoted. Get the basics right on the main pages, spend the real effort on the blog.

Do I need an llms.txt file?

No, and I'll save you the trouble: it's turned out to be decoration. The idea was a tidy file telling AI crawlers where your good content lives, a sort of robots.txt for language models. Lovely in theory. In practice, Ahrefs looked at whether engines actually fetch it and found that 97% of sites never had their llms.txt requested at all. It's a spec very few crawlers honour. By all means add one, it's five minutes, but don't mistake it for the work. The schema and structure are what get read.

The llms.txt file is lipstick. Almost nothing fetches it. Spend the hour on structure and references instead.

What's the single biggest lever?

Two levers, really, and neither is a clever file. The first is on-site structure: the answer-first, question-led, schema-marked content above, done consistently across the blog. The second is off-site references and backlinks: other credible sources mentioning and linking to you, which is what tells an engine you're worth trusting over the alternatives. On-site structure makes you readable; off-site references make you credible. You need both.

That off-site half is exactly why a technically perfect site can still go unquoted, which I get into in why your brand doesn't show up when buyers ask AI. Structure gets you to the door; references get you through it.

If you'd rather see it built than build it, our AEO, GEO and SEO work is the citation-ready version of all of this, and the same structural thinking underpins how we approach websites and conversion: a site legible to machines is usually clearer for humans too.

Want us to audit your current structure and hand you the checklist scored against your own site? Book a free Growth Audit and we'll show you what's readable, what's not, and what to fix first.