Microdata
Microdata is a markup format for embedding structured data straight into your existing HTML tags. Attributes like itemscope, itemtype and itemprop label individual pieces of content — a price, an address, an opening hour — so machines can tell what a piece of text means, not just that it exists. It's one of three ways to publish schema.org data on a page, alongside JSON-LD and RDFa.
Why microdata matters for AI visibility
Structured data of any kind, including microdata, makes it easier for crawlers and language models to parse what a page is actually saying — which field is a price, which is a rating, which is a business hour. That said, keep the claim honest: Google's own guidance on AI Overviews and AI Mode states plainly that no special markup, schema, or AI-specific file is required to appear in AI answers, and it explicitly warns against writing content "for AI" instead of for readers. So microdata isn't a citation hack or a ranking lever. What it does is remove ambiguity from content you've already written well, which lowers the odds of a system misreading or mangling your facts before it repeats them. That matters more than it sounds: an eight-tool study by the Columbia Journalism Review found AI search tools got basic sourcing details wrong more than 60% of the time. Clean, unambiguous markup won't fix hallucination, but it removes one easy source of it on your own pages.
How microdata works technically
Microdata uses three attributes layered onto tags you already have. itemscope opens a block that describes one thing — a product, a local business, an event. itemtype points that block at a vocabulary, almost always schema.org, which defines which properties are valid inside it. itemprop then labels each property inside the block, such as name, price or ratingValue. The attributes sit directly on the visible elements, so the markup and the content are interleaved in the same HTML. That's also its main drawback next to JSON-LD: because the labels are woven through the visible markup, the code gets harder to scan and easier to break when a template changes. Microdata, JSON-LD and RDFa are the three formats schema.org itself documents; all three describe the same vocabulary, just packaged differently.
Common mistakes with microdata
The most common error is marking up content that visitors never actually see — a five-star rating with no visible stars, hours that don't appear on the page. Search engines treat invisible structured data as a manipulation signal and can penalize the page for it. Next most common: broken nesting, where an itemprop sits outside the itemscope it's supposed to belong to and points at nothing. Third: itemtype values that are misspelled or don't exist on schema.org, which simply get ignored rather than flagged. Run your markup through Google's Rich Results Test or the Schema Markup Validator before you publish — both catch missing required fields and bad nesting before they cost you anything.
Microdata in relation to JSON-LD
Microdata predates JSON-LD and is now the less common choice for new work. Google and most AI systems favor JSON-LD because it lives in a single script block, separate from the visible HTML, which keeps templates cleaner and easier to update without touching the markup itself. Microdata is still valid and still gets read by crawlers and assistants — there's no need to rip out an existing implementation that works. For a new build, JSON-LD is the easier path to maintain. Either way, the format itself is not what determines whether an AI system cites you: what matters is that the structured data is accurate and matches the visible content exactly. Mismatched or stale structured data is worse than no structured data.
Example
Picture a small bike shop that does repairs and sales out of one storefront. The name, address and hours sit on the homepage as plain text a person reads without thinking. To a crawler, that's just a string of characters with no labeled meaning. The owner adds microdata around the existing text: itemprop="name" on the business name, itemprop="streetAddress" on the street, matching properties on the opening hours. Nothing on the page changes visually. But now, when someone asks an assistant "when does the bike shop on Elm Street open," the system has an unambiguous field to read the hours from instead of guessing at them from surrounding text.
Common questions
Are microdata and schema.org the same thing?
No. Schema.org is the vocabulary — the defined list of types and properties like Product or openingHours. Microdata is one of three formats used to attach that vocabulary to a page. The other two are JSON-LD and RDFa.
Does adding microdata get my page cited more by AI tools?
Not directly. Google says no special markup is required for AI Overviews or AI Mode, and structured data isn't a documented independent ranking or citation factor. What it does is make facts you've already published unambiguous to a machine, which reduces the chance they get misread or garbled.