Structured Data for AI Search | Make Your Content Machine-Readable

The bit everyone gets wrong about structured data

Right, so I've had this conversation maybe fifteen times in the last month. Business owner hears about structured data, maybe reads something about Schema markup, thinks "OK we need that for SEO" and then asks me what it actually does.

And look, fair question. Because the answer everyone gives is "it helps search engines understand your content better" which is... technically true but also completely bloody useless as an explanation.

Here's what's actually happening in 2026, and why it matters more now than it ever did before.

What structured data was for (and why that's changed)

Structured data, Schema markup, whatever you want to call it, used to be about getting those nice rich snippets in Google. Star ratings, FAQs, event details, recipe cards. Visual stuff that made your listing stand out.

That was the pitch. And it worked. You'd add some JSON-LD to your pages, Google would sometimes show your stars or your price range or whatever, you'd get a better click-through rate. Job done.

But that was when search engines were still basically just matching keywords and showing you ten blue links with some extras sprinkled on top.

Now? Perplexity, ChatGPT search, Google's AI Overviews, Claude with search, Gemini, all of them, they're not building a list of links. They're building an answer. And to build that answer, they need to pull facts out of your website and reassemble them into something coherent.

Which is exactly what structured data was designed for. Just nobody was really using it that way until now.

The AI doesn't read your site the way you think it does

Had a plumber in Baldock ask me last week why ChatGPT kept getting his service area wrong. He'd written it clearly on his homepage. "We cover Baldock, Letchworth, Hitchin and surrounding areas." Right there. Plain English.

But when someone asked ChatGPT "who does emergency plumbing in Letchworth" his business wasn't coming up.

So we looked at what he actually had in his markup. Nothing. No LocalBusiness schema. No areaServed properties. No service schema at all.

The AI had to guess from the text. And sometimes it guessed wrong. Or it just didn't bother because there were other businesses that had explicitly structured their service areas, and those were easier to extract.

That's the thing. AI search engines can read your content. Course they can. But when they're trying to answer a question quickly, they prioritise the information that's already been labelled for them.

It's like the difference between me asking you to find something in a box of random papers versus a filing cabinet. You'll find it eventually either way, but one of those is going to get answered first.

What it actually does in practice

Let me get specific because this is where it gets interesting.

Say you're a dentist in Stevenage. Someone asks Perplexity "which dentists in Stevenage do Invisalign for adults."

Without structured data, Perplexity has to: - Work out you're a dentist (probably fine, it's in your title) - Work out you're in Stevenage (probably fine if you mention it enough) - Work out you do Invisalign (depends if you have a page about it) - Work out you treat adults specifically (now it's guessing from context)

With proper schema, you've told it: - @type: Dentist (not ambiguous) - areaServed: Stevenage (explicit location) - hasOfferCatalog: Invisalign treatment (exact service) - audience: Adults (specific demographic)

And yeah, a smart AI can probably work most of that out anyway. But you know what it can't easily work out? Your prices. Your opening hours. Whether you take NHS patients. How long a typical appointment is. Whether you do emergency appointments.

All of that can be structured. And when it is, you're giving AI search engines exactly what they need to include you in answers.

The stuff that actually moves the needle

I'm going to give you the schema types that I've seen make the biggest difference for local businesses in North Hertfordshire. Not the exhaustive list, the practical one.

LocalBusiness and its subtypes. This is the foundation. Dentist, Plumber, Electrician, whatever. Your NAP (name, address, phone), opening hours, payment methods. Basic but critical.

Service schema. Each thing you do. Not just "plumbing services" but "boiler installation", "emergency leak repair", "bathroom fitting." Each one as a separate Service object with its own description, service area, typical price range if you can.

FAQPage schema. This one's huge. You know how AI Overviews and Perplexity love pulling FAQ content? That's not an accident. Questions and answers are already in the exact format they need. Mark them up properly and you're basically handing them quotable content on a plate.

Review schema. Star ratings still matter. Maybe more than before because AI search results often mention ratings when they recommend businesses. "Based on 47 reviews with an average of 4.8 stars" or whatever.

Breadcrumb schema. Helps them understand your site structure. Which sounds boring but actually matters when someone's asking about a specific service and the AI needs to work out if that page is about domestic or commercial work.

The Hitchin accountant who got this right

Client of mine, small accountancy firm in Hitchin. Two partners, handful of staff. They weren't ranking badly, they had decent content, but they weren't showing up in any AI search results.

We went through and added proper ProfessionalService schema for every service they offered. Not just "accountancy" but "small business accounts", "self-assessment tax returns", "VAT returns", "payroll services", each one structured with typical prices, who it's for, how long it takes.

Added FAQPage schema to their FAQ section (they already had good questions, just hadn't marked them up).

Added Organization schema with their accreditations, professional memberships, that sort of thing.

Three weeks later, started appearing in ChatGPT search results. Five weeks, showing up in Perplexity when people asked about accountants in North Hertfordshire. Not every time, but regularly enough to matter.

They tracked seven new client enquiries in two months that specifically mentioned finding them through an AI search. At their average client value, that's meaningful money.

And look, I can't prove it was just the structured data. They'd been doing other stuff too. But the timing lines up pretty well.

It's not instead of good content

Quick thing because I know someone's going to think this: structured data doesn't replace having actual good content on your site.

If your service pages are thin, if you haven't explained what you do properly, if your content's rubbish... schema won't save you. The AI still needs something worth including in its answer.

What schema does is make sure that good content gets understood and extracted correctly. It's an amplifier, not a replacement.

Had a kitchen fitter in Letchworth who wanted to skip writing proper service descriptions and just add schema. Doesn't work like that. You need both. The content for humans (and for AI to read), the schema for machines to parse efficiently.

The bit that takes actual work

Right, so here's where I'm going to be honest with you. Implementing structured data properly is a bit of a faff.

You can't just slap a plugin on and hope for the best. I mean, you can, and it might do something, but it won't do it well.

You need to actually think about: - Every service you offer and how to describe it - Your actual service areas (not just "Hertfordshire") - Realistic price ranges (ballpark is fine) - Your opening hours including exceptions - Which questions you get asked most often

Then you need to write it correctly. JSON-LD format. Nested properties. The right vocabulary from schema.org. Test it with Google's Rich Results Test and the Schema Markup Validator.

Get it wrong and it just gets ignored. Which isn't the end of the world, but it's wasted effort.

This is basically what we do with AEO in North Hertfordshire. Not just structured data, but the whole package of making sure local businesses show up properly in AI search results. Content, schema, the technical bits, monitoring what's actually working.

If you want to do it yourself, you absolutely can. Schema.org has documentation. It's not the most exciting reading but it's thorough.

Or if you'd rather just have someone sort it out properly, book a call and we can talk about what makes sense for your business. No hard sell, just a conversation about whether this is worth doing for you specifically.