Skip to content
Outrich Data

Glossary

Acceptance checks

Acceptance checks are the tests a delivered dataset must pass before the buyer accepts it and escrowed funds are released. Typical checks cover file integrity, record volume, schema, time span, de-identification and duplicates. Each check should name a property, a method, a pass threshold and the party who runs it.

Acceptance checks are agreed before signing and attached to the license as a schedule. Their purpose is to replace opinion with measurement. A check that says the data must be of good quality cannot be run. A check that says every line must parse as a JSON object with the listed keys can be run by anyone and gives the same answer each time.

Checks fall into two groups. Automated checks run across the whole delivery: hash comparison against the manifest, record counts, schema validation, pattern scans for identifiers and secrets, and duplicate rates. Manual checks rely on a sample, drawn at random with a recorded seed so either party can reproduce it. The schedule should say what follows a failure, including the notice the buyer must give and the seller's cure period. Properties that cannot be measured, such as general accuracy, belong in the warranties and not in the acceptance schedule.

Checks should be tried on the sample before the full delivery. If both parties run the agreed tools on the same sample and get the same results, the method is settled, and the final review repeats a process that already worked.

Example

A schedule might require that all files match the SHA-256 manifest, that the record count meets the agreed floor, and that a seeded random sample contains no direct identifiers.

Find out if your data qualifies.

Eleven questions, about five minutes, no call required. You get an answer on screen and a written follow-up.

esc