Skip to main content
Labeling operations fail quietly: throughput drifts down, rework creeps up, and nobody notices until the deadline does. The Performance tab turns those drifts into numbers you can check in thirty seconds a week.
PixlHub Performance tab with total tasks, quality rate, average label time, active contributors and throughput chart

The Performance tab: headline cards, the throughput trend against the previous period, and the quality-signal rates.

​
The headline cards

Throughput charts completed tasks per day against the previous period, so a “+40%” or a sagging line is visible before anyone writes a status report.

​
The quality signals

Three rates on the right tell you where quality really stands:
  • Rework Rate, tasks rejected at least once: the direct cost of unclear guidelines or rushed labeling.
  • Skip Rate, assignments skipped: rising skips usually mean confusing data or missing guideline coverage, not lazy annotators.
  • Reviewer Edit Rate, tasks modified by the reviewer: quick fixes in review are convenient, but a high rate means labelers never learn what reviewers keep correcting.
These three move together in tellable patterns: high rework with low reviewer edits means feedback loops work; low rework with high reviewer edits means reviewers silently absorb the quality gap. Aim for both low, with feedback flowing through rejections and comments.

​
Scope and share

The date range filter scopes every card and chart to the window you care about (sprint, week, delivery batch), and Export takes the numbers out for client reports or your own dashboards.

​
A weekly reading habit

  1. Throughput trend first: is the pace holding against last period?
  2. Quality Rate second: is first-pass yield stable while pace holds?
  3. If either slipped, the three rates plus the Tasks table localize it: which batch, which day, which step.
Five minutes, once a week, and the project never surprises you.

​
Frequently asked questions

Metrics reflect completed review cycles within the selected period; a task rejected and still in rework counts when its journey resolves. Widen the date range to see the full cycle.
Active Contributors counts participation in the period, and the Tasks table’s per-task ownership, duration and history provide the per-person view for deeper digging.

​
Next steps

Reviewing tasks

The process behind the quality numbers.

Tasks

The per-task detail when a metric needs explaining.