Okay, I need to talk about something that's going to make approximately zero people excited at a dinner party. Schema markup. I know. I know. But stay with me, because this boring little piece of behind-the-scenes code might be the single most important thing separating "websites AI systems can understand" from "websites AI systems politely ignore while recommending your competitor."

Here's the irony that's been bugging me for weeks: we've spent the last two years collectively losing our minds about AI search — how to get cited by ChatGPT, how to show up in AI Overviews, how to get Perplexity to whisper our brand name into the ears of potential customers. Meanwhile, most of us have been completely ignoring the mechanism that determines whether AI can understand our content in the first place.

It's like spending months perfecting your CV and then emailing it as a corrupted ZIP file.

Schema markup is essentially a set of invisible labels you stick on your website content — labels written in a language machines can read without having to guess what anything means.

Without schema, a search engine lands on your page and sees text, images, numbers, and has to play detective. It sees "£299" and has to figure out whether that's a price, a discount code, or some kind of abstract art reference. It sees "Open 9–5" and has to work out whether that's your business hours, a book title, or a surprisingly on-the-nose description of your work-life balance.

For traditional search, this inference game has worked reasonably well. For AI systems that need to extract, verify, and cite specific facts with precision? Not so much. Schema removes the guesswork entirely. It's code that explicitly tells machines: this is a product, this is its price, this is the brand, it's in stock, here are 2,100 customer reviews. The machine doesn't interpret. It just reads.

The format that's become the standard is JSON-LD, which sits in a separate block in your page's code. You can add it without touching anything your visitors see. No redesign, no layout chaos, no panicked Slack messages to your developer at 11 PM. It's a behind-the-scenes data layer that tells machines what your visible content actually means.

And despite all of this, only about 30 per cent of websites have implemented it at all. Seventy per cent of the internet is essentially speaking a language AI can't fluently read. Which is… a choice.

Before we get into the AI stuff, let's talk about the benefit schema has been delivering for over a decade that most people are still leaving on the table: rich results.

Those enhanced search listings showing star ratings, prices, availability, cooking times, FAQ dropdowns — all powered by schema. Without structured data, your listing shows up as a standard blue link with a text description. With it, you get visual prominence and extra information users can evaluate before clicking.

The numbers are not subtle. Rich results attract roughly 58 per cent of clicks versus 41 per cent for standard listings. When your competitor's listing shows a star rating, a price range, and a note about free delivery, and yours shows a title and two lines of grey text… well, I don't need to explain which one people are clicking. (Hint: it's the one that doesn't look like it was listed by a website from 2008.)

Quick misconception correction: schema is not a ranking factor. Google has confirmed it qualifies content for rich results but doesn't directly influence where you rank. The benefit is indirect — better click-through rates, stronger visual presence, clearer relevance signals. Which is still a lot of benefit for something most websites aren't bothering with.

Here's the thing: schema's role in traditional search was a nice-to-have. Its role in answer engine optimisation is foundational. And the reasons are pretty straightforward.

AI systems need to verify information before they cite it. When ChatGPT, Gemini, or Perplexity generates an answer that mentions your brand, it's not just pulling text off your page and hoping for the best. These systems cross-reference information across multiple sources to check consistency. Schema is one of the primary signals they check. If your Organisation schema says your company was founded in 2015 but your LinkedIn says 2016 and your Crunchbase says 2014, that kind of discrepancy — invisible to human visitors — is exactly the type of inconsistency that makes AI systems less confident about citing you. (Which means your dodgy "about us" page that hasn't been updated since the pandemic might be doing more damage than you think.)

Schema is how you become an entity, not just a website. Google's Knowledge Graph — the massive database that powers those information panels when you search for a company or person — is fed partly by structured data. When you implement Organisation schema declaring your company name, logo, founding date, and links to your profiles on LinkedIn, Wikipedia, and Crunchbase, you're not just labelling your content. You're telling Google's systems to treat your brand as a distinct, recognisable thing in the world. And when Google understands your brand as an entity, it surfaces more accurately in AI Overviews, brand panels, and AI-generated answers across its ecosystem. Analysis of over 107,000 websites cited in Google's AI Mode found Organisation schema on 82 per cent of them. That's not a coincidence.

Schema makes your content machine-citable. AI systems generate answers by extracting and synthesising information from sources. The easier you make that extraction, the more likely you are to be the source they pull from. Product schema that cleanly labels price, availability, specs, and reviews. Article schema that clearly identifies the author, publication date, and publisher. FAQ schema that pre-structures your content into the exact question-and-answer format AI systems are designed to work with. You're essentially pre-packaging your content in the format AI wants to consume it.

Schema must match your visible content — and that's actually the point. AI systems ignore schema that contains information with no visible counterpart on the page. You can't stuff hidden keywords or aspirational claims into your markup and expect AI to parrot them back. The schema has to confirm and structure what's already there. Which means this isn't a shortcut — it's a way of making your genuinely good content readable to the systems that increasingly decide who gets cited and who gets ignored.

One technical detail worth flagging: AI crawlers can't run JavaScript. If your schema is loaded dynamically through Google Tag Manager or client-side scripts rather than written directly into the page code, AI systems will never see it. This is worth checking with your developer, because it's a surprisingly common setup that effectively makes your structured data invisible to the exact systems you're trying to reach.

Here's the good news: implementing schema is considerably simpler than most people assume.

Check what you already have. Most modern CMS platforms — WordPress, Shopify, Wix, Webflow, Squarespace — generate some basic schema automatically. Google's Rich Results Test (free) lets you enter any URL and see what structured data exists. Run your homepage and key pages through it.

Use your platform's built-in tools. WordPress plugins like Yoast or Rank Math let you configure Organisation, Article, and Product schema through settings panels. Shopify generates Product schema automatically from listings. No code required for the most common types.

Validate everything. Non-negotiable. Google's Rich Results Test and Schema.org Validator check formatting and eligibility. If you've used an AI tool to generate schema, validation is especially important — AI-generated schema regularly contains errors in date formats, missing fields, and properties that don't quite match spec.

Know when to call a specialist. Basic schema on a standard CMS? Most marketing teams can handle it. Custom schema for complex site structures, headless CMS implementations, or scaling across thousands of pages? That's where an SEO consultant with structured data experience earns their fee.

Treat it as maintenance, not a one-off. Prices change. Hours shift. Team members leave. Schema that was accurate six months ago might be misleading today, and misleading schema is worse than no schema. A quarterly check of your highest-traffic pages is a reasonable minimum.

Schema markup is infrastructure, not strategy. It doesn't create authority, build trust, or generate original insight. It makes the authority, trust, and insight you already have legible to machines.

The 70 per cent of websites without schema aren't just missing rich results in traditional search. They're missing the basic infrastructure that determines whether AI systems can understand, verify, and cite their content at all.

And in an information environment increasingly mediated by machines, that's an expensive gap to leave open. Though I suppose if you're in the 30 per cent who've already sorted this, you can just sit back and enjoy the competitive advantage while the rest of the internet catches up. Which, knowing how quickly most organisations move on "unsexy infrastructure projects," should give you a comfortable head start of approximately forever.

Jo Lambadjieva is an entrepreneur and AI expert in the e-commerce industry. She is the founder and CEO of Amazing Wave, an agency specializing in AI-driven solutions for e-commerce businesses. With over 13 years of experience in digital marketing, agency work, and e-commerce, Joanna has established herself as a thought leader in integrating AI technologies for business growth.

😍It was great, all I needed

🙂Good, not great

🙄It sucked