Schema markup used to be a way to make your Google listing look nicer. Add the right code and you would get stars under your reviews, a price on your product, or an FAQ dropdown in the results. In 2026 its job has quietly and completely changed. Schema is now less about decorating your listing and more about telling AI who you are and what to trust, so your business gets named when someone asks.
If that sounds technical, stay with us. You do not need to write a line of code to understand this, and knowing which types matter lets you brief whoever builds your site so the effort goes where it counts.
What schema markup actually is
Schema markup is a small block of structured data added to a web page, usually in a format called JSON-LD, that describes what the page is about in a language machines understand. A human reads “Vioscale, Swindon, web design, 4.9 stars, open until 5pm” and just knows what it means. Schema spells the same facts out in a labelled way so a search engine or an AI does not have to guess: this is the business name, this is the location, this is the rating, this is the article, this is the author.
Think of it as a fact sheet stapled to the back of every page, written specifically for machines.
Why it matters more now, not less
Here is the shift that caught a lot of people out. After Google’s March 2026 core update, structured data stopped being mainly a trigger for those pretty search-result features and became a trust and entity-verification signal for AI. In plain terms: AI answer engines use your schema to confirm the facts about your business, resolve who you are against what they already know, and decide whether you are a safe source to quote.
And it is not only Google. JSON-LD is the shared format that Google’s AI Overviews and AI Mode, Microsoft Copilot, Perplexity and ChatGPT all read. Get it right once and every one of them can lift clean facts about you, rather than scraping them uncertainly from your page text and getting them wrong.
The numbers being reported back this up. Pages with complete, accurate structured data are cited noticeably more often in AI answers, with studies pointing to something in the region of a two to three times lift in AI citations for sites with solid schema versus those without. Google’s own AI Mode uses structured data for three jobs specifically: classifying what a page is, resolving your business as a known entity, and identifying which passage to quote.
The schema types that actually get you quoted
You do not need every type. For a small business, a short, well-chosen set does almost all the work. Here are the ones worth the effort, roughly in order.
- Organization (or LocalBusiness) schema, with sameAs. This is the single most valuable one now. It states your business name, logo, contact details and, crucially, your `sameAs` links, the URLs of your other verified profiles (LinkedIn, Facebook, your Google Business Profile, Wikidata if you have an entry). Those links let AI connect the dots and confirm “this is the same real business everywhere I look”, which is exactly what earns trust and a Knowledge Panel. If you only do one thing, do this.
- Article, BlogPosting or NewsArticle, with Author. For every blog post or guide, this schema tells AI what the content is, when it was published, and who wrote it. Named authors and clear dates feed the expertise signals AI weighs before quoting you.
- FAQPage. A well-marked FAQ remains close to ideal food for AI answers, because it pairs a real question with a self-contained answer, the exact shape a machine wants to lift. One change to know: Google now shows the old FAQ rich-result dropdown only on pages where the FAQ is the main content, so do not add it purely for the visual. Add it because AI reads it. We cover this in depth in FAQ and schema that AI loves to quote.
- Product and Offer. If you sell products online, this marks up price, availability and reviews, and still earns rich results as well as feeding AI shopping answers.
- Review and AggregateRating. Your rating and review count, made machine-readable. AI leans heavily on reputation when it decides who to recommend, so making this explicit helps.
- Speakable. A newer, more advanced one that flags the most quotable passage on a page. It has no visible search feature attached, but AI Mode uses it to pick what to synthesise, so it is worth knowing about for your key pages.
Notice what is not on the list. Desktop How-To rich results were removed, and marking up “editorial reviews” you wrote yourself can now trigger a manual penalty. The rule of thumb has not changed: only mark up what is genuinely on the page and true. Schema describes reality, it does not invent it.
What this looks like in practice
You do not add schema by hand on a modern site. If you are on WordPress, a good SEO plugin (or a properly built theme) outputs Organization, Article and FAQPage schema for you once it is configured. The work is in configuring it correctly and completely, not in typing code:
- Fill in your full business details and add every `sameAs` profile URL you own.
- Make sure each blog post has an author and a real publish date.
- Add genuine FAQs to your key pages and let the plugin mark them up.
- Keep the facts identical to what is on the visible page, and to what is on your Google Business Profile and other listings. Consistency is what makes AI trust the data.
- Validate it. Google’s Rich Results Test and the Schema.org validator will flag errors so nothing silently breaks.
If your current site outputs no structured data, or a tangle of half-configured tags, that is a common and fixable problem. It is also one of the higher-leverage jobs in AI visibility right now, because it makes every other signal you have easier for AI to read.
The bottom line
Schema markup is no longer about prettier search results. It is the clean, machine-readable fact sheet that tells AI who your business is, confirms it against the wider web, and makes you a safe source to quote. For a small business the winning set is short: Organization or LocalBusiness with `sameAs`, Article with an author, a real FAQPage, and Product or Review markup if they apply. Set those up accurately, keep them consistent everywhere, and you have done most of what structured data can do for your AI visibility.
Once the facts are readable, the next job is writing pages worth quoting, which we cover in how to write content that AI cites as a source, and understanding the AI-answer search you are optimising for in Google AI Mode explained.
Not sure what your site currently outputs, or whether AI can read your business correctly? That is exactly what we check. Take a look at our AI services, or book a free AI visibility chat and we will tell you what schema you have, what is missing, and the two or three fixes that will make the biggest difference.