Markdown

Subhub

What each industry can mark up, and what it actually does

Schema.org describes some industries in far more detail than others, and the size of an industry's vocabulary tells you very little about whether anybody uses it. I sorted all 937 classes into an industry by walking their parent types rather than by matching their names, then set what each industry can mark up against how much of it the web actually does. Pick your industry below and you get its own terms, the Google rich results that start from it, and the parts of it almost nobody has adopted yet.

More terms does not mean more adoption

July 2026 · six ranges, no exact numbersGoogle and the Schema.org community, Schema.org usage statistics dataset. Websites in Google's index.

Medical and health is the clearest case of that. It holds 88 types, more than any industry except media and publishing, and exactly one of them reaches a hundred thousand websites. Food and drink holds 14 types and two of those get there. Automotive holds 15 and not one does. So the length of an industry's vocabulary measures how much of it Schema.org has described, rather than how much of it anybody writes into a page.

Only the thirteen industries are drawn on the chart, and only they get a page of their own. The table carries two extra rows, because 333 classes are vocabulary machinery and 26 sit outside every industry, and both groups are named here rather than folded into whichever industry happens to be nearest.

Every industry, and how much of it the web uses. Adoption from Google's July 2026 usage file; the last two rows are not industries.
IndustryTypesReach 100,000 sitesUnder 1,000 sitesShare under 1,000Rich results
Media and publishing164326137.20%15
Medical and health8815764.77%0
Places6063761.67%0
Events and entertainment4821939.58%1
Local business, other454817.78%1
Retail and commerce436613.95%4
Travel and lodging3711437.84%1
Finance and legal2411041.67%0
Organizations, other223940.91%1
Automotive150320.00%0
Food and drink14217.14%1
Education143321.43%2
Jobs and careers4100.00%2
Vocabulary machinery, not an industry3333120862.46%5
Unassigned2621973.08%0

Four industries have no rich result at all

Google documents no rich result at all for four of the thirteen industries, and between them those four hold 187 vocabulary terms. Medical and health is the sharpest of them, with 88 types, one of them on a hundred thousand websites, and nothing Google documents a rich result for. Google documents rich results for what it chooses to show in search, and those choices are not spread evenly across what Schema.org describes. That does not make markup in those industries pointless, because a search result is only one of the things reading a page.

The industries Google documents no rich result for, with the vocabulary each one has.
IndustryTypesReach 100,000 sitesRich results
Medical and health8810
Places6060
Finance and legal2410
Automotive1500

Every industry

How this is measured

Two kinds of claim sit side by side on these pages and they are not equally strong. A type reaches its industry one way. By ancestry through rdfs:subClassOf, never by name. Each assignment records the ancestor that produced it. A Google rich result feature is a page on Google's website rather than a Schema.org type, so it has no parent to walk and it reaches its industry another way. One recorded judgement per feature, its primary Schema.org type, then assigned by the same ancestry walk. You can check every one of those judgements against Google's own documentation, because each feature carries the address of the page it came from.

That walk reads an ordered list of industries, most specific first, because LocalBusiness is a subclass of both Organization and Place and the three cannot all be industries. A type takes the first industry it matches on the way up, so it lands on one page rather than on three. Every assignment keeps the ancestor that won, and that ancestor is shown beside the term on the industry pages.

Not all of the vocabulary belongs to an industry at all. 333 of the 937 classes describe how markup is structured rather than what a business does, and everything under Action, Intangible or DataType is counted there. Another 26 sit outside every industry, mostly datatypes, units of measurement and the bioscience terms. Folding either group into the nearest industry would inflate every industry with terms belonging to none of them, so both keep their own names.

Google rewrites its own documentation without announcing it, so every feature carries the address of the page I read it from and that page's own last-updated date. I read them on 11 August 2026 and the dates on those pages run from 10 December 2025 to 24 July 2026. My recommended gap report stops at 7 July 2026, which is earlier, and Review snippet is the whole of the difference. That report lists only the features with at least one thinly used recommended property, and this one has none, so it appears here and not there.

Every adoption figure here is a band rather than a count. Google sorts each term into one of six and publishes nothing inside them, so a term you can see on somewhere between one and ten million websites has no public exact number, on this page or anywhere else. A longer list of types does not mean a bigger industry in the economy either, because these figures count vocabulary terms rather than businesses or websites.

You can download the same figures as JSON or CSV, and every term in the table on each industry page is in the explorer data with the same band.

Snapshot July 2026. 3 months archived. Built 18 August 2026.

Analysis © Eduard Dziak, licensed under the Apache License 2.0. Please credit eduarddziak.com with a link. Source data: Google and the Schema.org community, Schema.org usage statistics dataset. Rich result requirements: Google Search Central, licensed CC BY 4.0. Term descriptions: Schema.org, licensed CC BY-SA 3.0, each term linking to its own page. Crawl counts: HTTP Archive Web Almanac, licensed Apache License 2.0.