Which Schema.org Types Are Actually Used Across the Web? July 2026 Data

What the dataset measures

Schema.org and Google publish monthly files with three core fields: term class, term URI, and a bucket representing how many unique domains use the term. Schema.org says buckets reduce crawl noise and protect privacy. The files reflect the web as indexed by Google, exclude sites unavailable to that crawl such as robots-blocked sites, and do not distinguish markup formats.

We downloaded the official JSON files named 2026_05.json, 2026_06.json, and 2026_07.json on August 10. We matched exact Schema.org URIs and compared the published bucket strings. No interpolation was used, and a bucket change is reported only when the official label changed.

Selected publishing and commerce types

The table shows ranges, not the magnitude of the movement inside a range. A term at 9.9 million and a term at 1.1 million would share one bucket. A crossing into 10M+ proves only that the release assigned the term to the higher range.

The long tail dominates the vocabulary

Each monthly file contained 958 item types and 4,587 predicates. In July, 470 item types and 3,747 predicates were in the < 1K bucket. Only 16 item types and 36 predicates were in 10M+.

That imbalance is expected in a broad vocabulary. Specialized medical, government, scientific, and workflow terms address smaller populations of sites. Schema.org explicitly warns that a low bucket does not mean search engines ignore a term.

How publishers should use the data

• Choose types by visible meaning. Mark up the entity and content actually present, not the most popular term.
• Separate vocabulary adoption from search features. Google’s feature documentation defines additional requirements and supported properties.
• Watch stable changes. A bucket crossing is useful for ecosystem research, but monthly noise and crawl coverage still constrain the finding.
• Keep exact versions. Save the raw file, retrieval date, query, selected URIs, and derived table.
• Do not infer format. The data combines JSON-LD, Microdata, and RDFa at the domain level.

The site’s article comments and discussion schema guide shows why meaning and page structure matter more than a popularity contest. Use the evidence-led publishing guide to keep the raw dataset beside every derived claim.

Primary data and method

• Schema.org: Announcing the usage statistics dataset
• Schema.org: About usage statistics
• Official Google-contributed monthly files

Reproducibility note: the three raw JSON files and exact selected-term output were retained with this publication package. The analysis compares official bucket labels and does not estimate hidden counts.