Does schema markup help with AI search?
Short answerSchema markup will not get you into AI answers on its own. Google says AI Overviews and AI Mode need no special schema, and OpenAI and Perplexity publish nothing on whether they read it. It is still worth adding a few standard types, because it labels your key facts unambiguously for search engines whose results the AI tools draw on, and it costs little.
Key points
- No platform documents schema as a requirement or a ranking boost for AI answers; treat claims otherwise with suspicion.
- Organization or LocalBusiness, Service, Person and, for shops, Product cover what most businesses need.
- Markup must match what the page visibly says, or it can count against you.
- Google stopped showing FAQ rich results in May 2026, but FAQPage markup is still valid and the visible questions and answers still matter.
Schema markup is one of the first things sold as an "AI search fix", often at a price that implies it is the secret ingredient. It is not a secret and it is not the ingredient. It is a sensible, inexpensive piece of housekeeping that removes doubt about your business facts, and it is worth doing properly once you know its limits.
What schema markup is
Schema markup, also called structured data, is a short block of code on a page that labels facts for machines: this is the business name, this is the phone number, this is the address, this is the year it was founded. The labels come from schema.org, a shared vocabulary that Google, Microsoft, Yahoo and Yandex set up together.
Visitors never see it. It usually sits in the page's code in a format called JSON-LD, a small block of labelled text that your developer, theme or plugin adds. A tiny piece looks like this: "@type": "LocalBusiness", "name": "Hartwell Joinery", "foundingDate": "2009".
The point is to remove guesswork. On a normal page, a machine has to work out which number is your phone and which is your company registration. With markup, it is told.
What the platforms actually say
This is where most sales pitches go quiet, so here is what each platform documents.
- Google: its guidance on AI features says there is no special schema.org structured data you need to add to appear in AI Overviews or AI Mode. Its wider guide to AI in Search repeats that structured data is not required for generative AI search. Google does say structured data helps it understand pages, and that markup must match the visible content.
- Microsoft: Microsoft's guidance on appearing in AI answers says schema helps search engines and AI systems understand your content, and points to schema.org for the types that apply.
- OpenAI and Perplexity: their crawler documentation covers which bots to allow, and says nothing about whether ChatGPT or Perplexity read structured data or give it weight.
No platform says schema markup will get you named in AI answers. Anyone who promises that is selling a guess.
Why it is still worth doing
If it is not required, why add it? Four practical reasons.
- The search engines read it. Google and Bing use structured data to understand pages, and AI answers lean heavily on their results. ChatGPT search uses third-party providers including Bing, and Google's AI features sit on its core ranking systems.
- It settles ambiguity. If your business name is also a common word, or shares a name with a company elsewhere, Organization markup with your address, logo and social profiles helps Google tell the two apart. Google's own documentation describes some properties as existing specifically to disambiguate organisations.
- It still powers some Google features. Product prices and availability, review stars on products, breadcrumbs, events and your logo in results all rely on it.
- It is cheap. On most modern platforms it is a plugin setting or a few lines of code, done once and updated only when your details change.
The types worth adding
Most businesses need four or five types, not forty.
| Type | What it labels | Where it goes |
|---|---|---|
Organization or LocalBusiness | Name, logo, address, phone, opening hours, founding date, links to your profiles elsewhere (sameAs) | Homepage, or your about or contact page |
Service | Each service you offer, the area you serve, who provides it | Each service page |
Person | Founder or key experts: name, role, qualifications, profiles | About or team page |
Product with Offer | Price, availability, reviews of the product | Each product page (shops only) |
BreadcrumbList | Where the page sits in your site | Every page, usually automatic |
FAQPage | Questions and answers shown on the page | Pages with a visible FAQ section |
Choose the most specific business type that fits. schema.org has dozens of LocalBusiness subtypes, such as Dentist, AccountingService or HomeAndConstructionBusiness, and a specific type says more than a generic one.
A note on FAQPage: Google stopped showing FAQ rich results in its search results from May 2026, having already limited them to government and health sites in 2023. The markup is still a valid schema.org type, and Google has not asked sites to remove it. What matters more is the visible list of real questions and direct answers, which people and AI tools can both read.
A worked example
Sample business: Hartwell Joinery, bespoke kitchens in Leeds. The homepage carries one HomeAndConstructionBusiness block with:
name: Hartwell Joinery, exactly as on its Google Business Profile.addressandtelephone: the current Kirkstall workshop, not the old one.foundingDate: 2009.areaServed: Leeds and Harrogate.sameAs: links to its Google Business Profile, trade association listing, Instagram and Houzz profile.founder: aPersonentry for the owner, matching the about page.
Each service page (bespoke kitchens, utility rooms, fitted furniture) has a Service block naming the service, the area and Hartwell as provider. Every fact in the markup appears in plain text on the same page.
Notice what the markup does not include: the 4.9 rating from 128 reviews. Google's rules say that when a business controls the reviews about itself, its LocalBusiness or Organization pages are not eligible for review stars. Those reviews do their work on Google and the review platforms themselves.
Mistakes that do harm
- Markup that does not match the page. Google's structured data guidelines require markup to describe what visitors can see. Hidden or misleading markup can lead to a manual action, where Google removes rich results from your site.
- Two plugins, two answers. A theme and an SEO plugin each output an Organization block with different details. Now the code contradicts itself.
- Stale facts. You moved, changed your number or dropped a service, and the markup still says otherwise. Update it whenever you update your listings.
- Keyword stuffing in descriptions. A
descriptionfield crammed with search terms reads as spam to people and machines. - Self-written review stars. Marking up your own testimonials as ratings breaks Google's rules and can cost you more than it gains.
How to add it and test it
- Check what you already have. Run your homepage through Google's Rich Results Test and the Schema Markup Validator. Many sites already output some markup through their theme.
- Use your platform's tools. On WordPress, an established SEO plugin can generate Organization and breadcrumb markup. Shopify themes usually output Product markup. Custom sites need a developer to add JSON-LD blocks.
- Fill in every property that applies. Google recommends including as many relevant properties as you can, rather than the bare minimum.
- Watch Search Console. Its enhancement reports flag markup errors after Google recrawls your pages.
- Keep it in your redesign plan. Markup is easily lost when a site is rebuilt. The website migration SEO checklist includes it.
Where it sits in your priorities
Schema comes after the work that matters more: letting the search crawlers in, writing pages that say plainly what you do, and making your business facts match everywhere. If ChatGPT is naming a competitor, the guide to why ChatGPT recommends your competitor will usually point to bigger gaps first, and the guide to appearing in Google's AI Overviews covers Google's side. Treat llms.txt the same way: the guide to what llms.txt is explains why it is housekeeping too.
Structured data is included as standard in an M/AFZAL Rebuild, alongside a question-led FAQ, because it is far easier to get right when a site is built than to retrofit. For how it fits with everything else, see the main guide to AI search visibility.
Straight answers.
The follow-up questions owners ask most.
- Should I remove FAQPage markup now that Google has dropped FAQ rich results?
- There is no urgent need. FAQPage remains a valid schema.org type, Google has not asked sites to remove it, and other systems may still read it. Just do not expect a visual feature in Google from it.
- Is JSON-LD better than microdata?
- Google recommends JSON-LD where your site allows it, because it sits in one block separate from the visible page and is easier to maintain. Microdata and RDFa are also supported.
- Can my web developer add schema in an afternoon?
- For a small site with a few standard types, often yes, especially on WordPress or Shopify where plugins and themes do much of it. Larger sites, or sites with conflicting plugins, take longer to tidy up properly.
- Does Bing or Copilot use structured data?
- Microsoft says schema helps search engines and AI systems understand your content, and recommends schema.org as the reference. It has not published how much weight it carries in Copilot's answers.