Define the task
Each category starts with a repeatable job that buyers actually compare tools for.
Benchmarks
Benchmark categories
Yardstick Lab is organized by tool category. A category goes live only when there is a fixed task set, visible scoring method, and dated run behind it.
Live now
Document AI
The first live category compares tools that extract structured data from real documents.
Live benchmark
Document AI tool rankings
Current coverage includes structured extraction, tables and repeated rows, reconciliation-heavy documents, and hard language or layout cases.
50documents
6scored rows
5score components
Category rules
How new categories get added
The site stays broad by structure, not by pretending to have categories that have not been tested yet.
Freeze the set
Tools are tested against the same inputs, requested outputs, and scoring rules.
Publish a dated run
Rankings need a run date so later product changes create a new result.
Add reviews after evidence
Product reviews come after enough outputs exist to explain fit and limits.
Not ranked yet
Future categories stay unlisted until tested
Yardstick Lab can expand beyond document AI, but empty category pages would weaken the premise. The public site shows only tested categories.
No other benchmark category is live yet
New category pages appear only after the task, methodology, and first run are ready to publish. Until then, the broader Yardstick Lab shell lives here and the evidence lives in the Document AI category.