Skip to main content

​
QA workload management

Quality management gives you scorecards, calibration sessions, and a continuous auto-sampler (every 15 minutes) that drops calls into a shared queue. Workload management sits on top of that: instead of leaving every review up for grabs, a supervisor can say who reviews what, cap how much unfinished work any one evaluator can hold, and see who is falling behind — before it becomes a backlog. It works from the Quality → Evaluations → Workload tab in the dashboard and through the API. Everything here is reviewer-scoped: organization members with an owner or admin role. Agents never see the workload view; they only see their own scores.

​
Assign a call to a specific evaluator

From the dashboard, open Quality → Evaluations → Workload and click Assign evaluation — pick an active scorecard and the evaluator, and the assignment lands as pending on their queue. Over the API:
The assignment is open — pending and unscored — until the evaluator submits real scores through the normal grading flow. Two assignments for the same call return 409 ALREADY_ASSIGNED; the continuous auto-sampler and CSAT-triggered reviews coexist with manual ones and can claim the same scoring form afterward.

​
Rebalance an open assignment

Any pending, unscored assignment can be moved to a different evaluator — from the Reassign action on the Evaluations tab, or via the API:
Once a score is submitted the assignment locks: reassigning it returns 409 INVALID_STATE, because a scored review has an audit trail an evaluator owns. Self-evaluations and system-generated rows (the auto-sampler’s unclaimed stubs, CSAT-triggered and AI auto-scored rows) are never reassignable.

​
Per-evaluator quotas

Each evaluator has a cap on open (assigned-but-unscored) work — 25 by default. Assigning to an evaluator already at cap returns 409 QUOTA_EXCEEDED with the server’s message, and the dashboard picker pre-disables full evaluators so you find out before submitting. Tune the cap per organization through your workspace settings (the General settings surface) under qa_evaluation_assignment.max_open_per_evaluator; the change applies immediately to every existing and future assignment.

​
Due dates

Every assignment carries a due date: its creation time plus your review window — 3 days by default, tunable per organization as qa_evaluation_assignment.due_days. The due date is derived on every read rather than stored, so changing the window re-buckets overdue counts immediately and you never backfill a column.

​
See everyone’s workload

The Workload tab lists every eligible evaluator with their open count against the quota, remaining capacity, and how many of their open assignments have passed the due-date window. Over the API:
The response carries the tenant’s resolved due_days and max_open_per_evaluator plus an items[] row per evaluator (open_count, overdue_count, quota, remaining). Add ?evaluator_id=user_... to narrow to one evaluator.

​
Endpoints

​
See also

  • Quality API reference — scores, calibration sessions, and the scorecard pipeline these assignments feed into
  • Team chat — coordinate a review queue rebalance with your QA leads