Structured Data Tester for AI Search

Paste a JSON-LD block below and check it for valid syntax plus the required and recommended fields that AI search engines - ChatGPT, Perplexity and Google AI Overviews - rely on to identify and cite your pages. Everything runs in your browser. Nothing is sent to a server.

Google retired its own Structured Data Testing Tool. This checks the schema types that matter for AI citation: Product, FAQPage, Organization, BlogPosting, Article, BreadcrumbList and more.

Why AI engines read your structured data first

Here is the thing most stores miss. When ChatGPT, Perplexity or Google AI Overviews build an answer, they do not read your page the way a shopper does. They parse it. JSON-LD is the cleanest signal you can hand them - a machine-readable statement of what the page is, what you sell, what it costs and who stands behind it.

Get that block right and you are easy to extract and easy to cite. Get it wrong, or skip it, and the engine guesses from your visible text. That is where brands quietly disappear from AI answers.

From auditing 50+ brands, the pattern is blunt: the stores that show up in AI answers almost always have valid, complete schema on their key pages. The ones that do not are usually missing it, or shipping markup with errors that silently drop the page from consideration.

How to use this tester

1. Grab your JSON-LD. View source on the page, or open DevTools and search for application/ld+json. Copy everything inside the script tag.
2. Paste and validate. Drop it in the box above and hit Validate schema. You can paste the whole <script> tag - the tool strips it for you.
3. Fix the red lines first. Errors are missing required fields and they break the entity. Warnings are recommended fields - they improve extraction but will not block a citation on their own.
4. Re-test until clean. Then confirm the block actually lives in the page source, not injected so late that a crawler never sees it.

What each schema type needs to get cited

The tester enforces the fields below. This is the short version of what AI engines and Google's rich results actually look for.

Product

Required: name, plus at least one of offers, review or aggregateRating. Ship image, description, brand and a full offers block with price, priceCurrency and availability. Price and availability are exactly what an AI shopping answer quotes. Leave them out and you are a name with no reason to be recommended.

FAQPage

Required: mainEntity - an array of Question nodes, each with a name and an acceptedAnswer that has real text. The text has to match an answer a human can see on the page. Schema that does not mirror the visible page is the fastest way to get your markup ignored.

Organization

Required: name and url. Add logo, sameAs (your real social and Wikipedia or Wikidata profiles) and contactPoint. This is your brand entity - how an engine confirms the company behind the products is a real, trusted thing.

BlogPosting / Article

Required: headline, image, author and datePublished. Add dateModified and a publisher. A named author object beats a bare string every time - E-E-A-T is a citation signal, not a checkbox.

BreadcrumbList

Required: itemListElement. Cheap to add, and it tells an engine where the page sits in your catalogue - useful context when it picks which page to surface.

The 5 schema mistakes I see most

1. Markup that contradicts the page. A price in JSON-LD that does not match the one on screen. Engines trust the page, then distrust your schema.
2. FAQ schema with no visible FAQ. Google walked back FAQ rich results for most sites partly because of this. Keep the answers on the page.
3. Product with no offers. No price, no availability - barely a product to an AI shopping answer.
4. Author as a plain string. Use an author object with a name. It is the difference between anonymous and attributable.
5. Broken JSON. One trailing comma and the whole block goes invisible. That is what the syntax check above is for - run it before you ship.

Frequently asked questions

Is this structured data tester free?

Yes. Paste your JSON-LD and validate it in the browser - no login, no limits, no API calls. Nothing you paste leaves your machine.

What schema types does it check?

It validates JSON-LD syntax and checks required and recommended fields for Product, FAQPage, Organization, BlogPosting, Article and BreadcrumbList - the types AI search engines use to understand and cite e-commerce pages.

Why does AI search need structured data?

ChatGPT, Perplexity and Google AI Overviews parse JSON-LD to identify products, prices, FAQs and brand entities. Valid, complete schema makes a page easier to extract and cite. Missing or broken schema forces the engine to guess from raw text, and guesses do not get cited.

Does valid schema guarantee I will get cited?

No. It removes a blocker, it does not buy you a ranking. Schema makes you eligible and easy to parse. Whether you actually get cited depends on authority, content depth and how often the engines crawl you. The audit below shows where you stand on all three.