Acceptance Rate
The percentage of an annotator's submitted work that is approved by review.
Acceptance rate is the single most visible quality metric on most annotation platforms: the share of submitted tasks that pass review without rejection or rework. It's a lagging, aggregate signal — it tells you how often you're getting things right, but not why, which is where error analysis and rejection reason codes come in.
Most platforms gate access to higher-paying or more specialized task queues behind a minimum acceptance rate, which makes this the number annotators tend to track most closely day to day.
Because acceptance rate is downstream of reviewer judgment, it's only as reliable as the reviewers applying it — see reviewer consistency for the flip side of this metric.
What this means for trainers
A dropping acceptance rate with no clear explanation is worth investigating quickly — it can mean a guideline changed without clear notice, or it can mean a specific reviewer is applying standards differently than the rest.
Related terms
Put this into practice
Browse open AI training roles from Alignerr, Mercor, Outlier, and more.
Browse AI training jobs