Every article about getting recommended by AI — including ours — eventually mentions "structured data" or "schema markup," usually in a tone that assumes you know what that means. Let's fix that properly, with zero code required to follow along.
The jam jar analogy
Imagine two pantries. In the first, every jar is unlabelled — to find the apricot jam you open jars and taste. In the second, every jar has a printed label: contents, date made, ingredients. Same jam. One pantry is usable by anyone instantly; the other requires effort and guesswork.
Your website is a pantry, and machines are the visitors in a hurry. A human reading your page uses judgement to work out that "Est. 1994 in the heart of Paarl" means you're a business in Paarl founded in 1994. A machine parsing millions of pages can't afford that judgement on every jar. Schema markup is the printed label: a hidden layer on your pages that states, in a standard format machines agreed on, exactly what each fact is.
The label doesn't change how your website looks. Visitors see your beautiful pages exactly as before. The labels sit underneath, readable only by the systems deciding whether to show, rank — and now recommend — you.
What a label actually says
Without getting technical, a schema label on a guesthouse's homepage effectively tells machines:
This page describes a guesthouse • name: Vine & Valley Guesthouse • location: Paarl, Western Cape, South Africa • phone: +27 21 … • price range: R1,450–R2,200 • amenities: pool, parking, pet-friendly rooms • rating: 4.8 from 214 reviews • hours: …
Compare that with a machine trying to deduce the same facts from flowing marketing prose, and you see the point: schema removes ambiguity. And ambiguity is exactly what stops cautious systems from recommending a business — confidence is what recommendations are made of.
Why this suddenly matters more
Schema has helped with Google for years (it powers the star ratings, prices and FAQ dropdowns you see in search results). But the stakes rose with AI answers. When ChatGPT, Perplexity or Google's AI reads your site while composing a recommendation, pages with clear structured labels are simply easier to extract correct facts from — your prices, your location, your offerings, stated unambiguously in a format built for machine reading. Easier to understand correctly means safer to recommend. It's the same principle as everything else in GEO: reduce the machine's uncertainty about you, and you get chosen more.
The label types that matter for a local business
Schema has hundreds of types; a small business needs a handful:
- LocalBusiness (or its specific versions — restaurant, lodging business, plumber, and so on): the master label with your name, address, phone, hours, area served and price range. If you add only one, add this.
- Service or Product labels: one per thing you sell, so "geyser replacement" exists as a labelled fact rather than a phrase in a paragraph.
- FAQ labels: your question-and-answer content, marked as such — a natural pairing with the seven questions your site should answer.
- Review/rating labels: letting your review reputation travel with your pages.
- Organization: who's behind the site, linking your website to your social profiles and Google presence so machines connect the dots into one confident picture of one business.
How to check whether your site has any (2 minutes, no code)
Two easy checks. First, paste your homepage address into Google's Rich Results Test (search that phrase — it's a free Google tool) and see what structured data it detects. Second, try the validator at schema.org the same way. Green results with detected items: you have labels — check the details are correct and current, because a wrong label (old prices, old address) is worse than none. Nothing detected — very common among the local sites we audit — means machines are back to tasting unlabelled jars.
Do you have to touch code to fix it?
Often, no. Most modern website builders (WordPress via popular SEO plugins, Wix, Squarespace and others) can generate basic business schema from a settings form — check your platform's settings or your web person's scope of work; if they built your site recently, ask them plainly: "does our site have LocalBusiness schema, and is it current?" It's a fair question and a five-minute check for them.
The honest caveat: getting beyond the basics — specific business types, per-service labels, FAQ markup that matches your actual content, keeping it all correct as your prices and offerings change — is fiddly, and mistakes are invisible to you while being obvious to machines. This is precisely the unglamorous plumbing our Quick Boost package fixes in week one, and it's on our own Client Zero to-do list for exactly the reasons above.
The takeaway
Schema markup isn't a magic ranking trick, and anyone selling it as one should be avoided. It's something humbler and more important: making sure that when machines read about your business, they understand it — correctly, completely and confidently. In an era where machines summarise you to your customers, that understanding is the ground floor of being recommended.
Wondering whether your site is labelled, mislabelled or a pantry full of mystery jars? It's one of the first things we check.