Most small business owners think schema markup died when the star ratings vanished from their search listings. It did not. It changed jobs. The visual extras got quieter, and at the same time your markup became the cleanest way for Google and the AI engines to understand what your business is, where it is, and whether to name it as a source.
That is a bigger deal now than it was two years ago. Here is what structured data actually is, why 2026 changed the math on it, the schema types still worth your time, how it gets built, and how to tell if it is working.
What structured data actually is, in one minute
Structured data is code that labels the meaning of things on a page. A machine reading your homepage sees a wall of text and images. Schema markup adds hidden labels that say "this string is the business name," "this is the phone number," "this is a $$ price range," "this block is five customer reviews averaging 4.8."
The format almost everyone uses is JSON-LD: a single block of code in the page that describes the facts, without touching how the page looks to a visitor. You are not writing content twice. You are handing the search and AI engines a clean summary of what the page already says.
Why 2026 changed the math
For years, the payoff for schema was visual: rich results. Star ratings, FAQ dropdowns, recipe cards, and how-to steps that made your listing bigger in Google. Then Google pulled back. It has wound down FAQ and HowTo rich results in Search for most sites, so the old "add FAQ schema, get a bigger listing" trick largely stopped paying out.
If that were the whole story, schema would be fading. It is not, because a second, larger use arrived. AI answers now sit on top of search. By third-party estimates, Google's AI Overviews show on roughly half of US searches, and tools like ChatGPT and Perplexity answer millions of buyer questions a day. Those engines lean on structured, machine-readable signals to decide what a page is about and who to trust.
Schema stopped being a trick for a prettier listing and became the label that tells an AI engine what your business is and why to cite it.
The evidence is directional, not magical: one 2026 analysis found that around two-thirds of the pages ChatGPT cites include structured data. It is not the only factor, and it will not carry thin content. But when an engine is choosing between two similar pages, the one that is clearly labelled is easier to read, easier to trust, and easier to quote. Being uncited is the new invisibility, and structured data is one of the cheaper ways to stay legible.
The schema types that earn their keep
You do not need the whole schema.org vocabulary. For most small businesses, a short list does the heavy lifting:
- Organization. Defines your brand as an entity: name, logo, official URL, social profiles. This is the anchor that helps engines recognise you consistently across the web.
- LocalBusiness. Name, address, phone, opening hours, price range, service area. This is what feeds "near me" answers and local knowledge panels. If you have physical locations, this one is not optional.
- Product or Service. For what you sell: name, description, price, availability. Essential for stores, useful for service businesses that want their offers understood.
- Article or BlogPosting. On your content, with author and publish date. It tells engines who wrote a piece and when, which supports both ranking and AI citation.
- Review or AggregateRating. Where you have genuine ratings on the page. Real social proof, made machine-readable.
Add BreadcrumbList if your site has real hierarchy, since it helps engines map how your pages relate. Everything past this list is a refinement, not a requirement. Start with the types that describe your actual business and stop there.
How it actually gets built
The good news: adding core schema to a small site is usually a half-day job, not a rebuild. The pattern is the same every time.
- List your facts. Business name, address, phone, hours, the services or products you want understood, and any real ratings.
- Write one JSON-LD block per type, placed in the page code. One source of truth, so the markup and the visible page never disagree.
- Match it to the page. Only mark up what a visitor can actually see. Schema is a label for real content, not a place to invent credibility.
- Validate before you ship. Run it through Google's Rich Results Test and the Schema Markup Validator until it parses with zero errors.
Here is what that clean, validated block looks like in practice, and why both Google and an AI engine can read the same markup:
What a clean schema block looks like in practice
One JSON-LD block, validated once, then read by both Google and the AI engines.
The point of that middle step is the whole discipline. Validated once, the block is then read by every engine that visits, and it says the same thing to all of them.
The mistakes that quietly break it
Schema fails in boring, avoidable ways. The four we see most:
- Markup that lies. Review stars or FAQs in the code that are not on the page. This breaks Google's guidelines and can earn a manual action that strips your rich results. Truthful markup only.
- Two plugins, two blocks. A theme and an SEO plugin both output Organization schema with slightly different details. The engine sees a contradiction and trusts none of it.
- Set and forget. You change your phone number or hours on the page but not in the schema. Now your markup is confidently wrong, which is worse than having none.
- Marking up nothing useful. Reams of generic WebPage schema that adds no facts an engine cares about. Volume is not the goal; accuracy about your actual business is.
Every one of these traces back to the same rule: the schema must describe what a human sees, from a single source, kept current.
How to tell if it is working
Structured data is measurable, so treat it that way. Confirm the code first: the Rich Results Test and Schema Markup Validator tell you it parses. Then watch Google Search Console's enhancement reports, which flag valid items and warnings for the types you added, and show you problems as they appear.
The AI side needs a different check. Ask ChatGPT, Perplexity, and Google's AI the exact questions your customers ask, and note whether your business gets named. If a competitor with clean markup shows up and you do not, that gap is your to-do list. Track "am I getting cited" as its own signal, separate from rankings and traffic, because in an answer-first world it is often the earliest sign you are being read at all.
Where this fits
Structured data is not a growth channel on its own. It is the plumbing that makes your other work legible: it helps your SEO in 2026 by clarifying entities and content, it feeds the citation game covered in our GEO and AEO guide, and it is a direct lever for winning the local map pack. Done right, one set of clean markup pays into all three at once.
We build validated schema into every site we ship and retrofit it onto sites we did not, as part of technical SEO. See the work we have done for clients, explore our SEO setup and SEO audit services, or tell us about your site and we will check what your markup is saying about you right now. Often it is saying nothing, and that silence is costing you the citation before the content ever gets a look.
FAQ
Questions, answered.
The structured-data questions clients ask before they commit.
Schema markup is a small block of code you add to a page that labels what things are: this is a business, this is its phone number, this is a product, this is a review score. Search engines and AI engines read your page as text, and schema removes the guesswork by telling them exactly what each piece means. The most common format is JSON-LD, a tidy block that sits in the page's code without changing how the page looks to visitors.
Yes, but the reason shifted. It used to be mostly about winning visual extras in search results (star ratings, FAQ dropdowns). Google has since wound down several of those rich result types. What grew in importance is machine-readability: schema helps Google and AI engines like ChatGPT, Perplexity and Google AI Overviews understand and trust what your business is, which feeds both rankings and whether you get named as a source. Schema will not rescue weak content, but it removes friction for good content.
For most small businesses, five cover the majority of the value: Organization (who you are as a brand), LocalBusiness (name, address, phone, hours, price range), Product or Service where relevant, Article or BlogPosting on your content, and Review or AggregateRating where you have real ratings. Add BreadcrumbList for site structure if it fits. You do not need every type in the schema.org vocabulary; you need the handful that describe your actual business.
Not always. Site builders and SEO plugins can generate basic Organization and LocalBusiness schema from a form. The trouble starts when the markup drifts from what the page says, gets duplicated by two plugins, or claims things (like review stars) the page does not actually show. That is where a developer or a studio earns their fee: making the schema accurate, single-sourced, and validated, so it helps instead of triggering a penalty.
It can, if it lies. Marking up reviews, prices, or FAQs that are not genuinely on the page is against Google's guidelines and can lead to a manual action that removes your rich results. The rule is simple: schema must describe what a human actually sees on the page. Keep it truthful and matched to visible content and there is no risk; use it to fake credibility and there is.
Use Google's Rich Results Test and the Schema Markup Validator to confirm the code parses with no errors. Then watch Search Console's enhancement reports for valid items and warnings over time. To see the AI side, ask the engines the questions your customers ask and note whether you get named. Validation proves the code is correct; Search Console and AI spot-checks prove it is doing something.
