Skills Quality assurance
Reviewer workload and grading quality
QA programmes often staff reviewers like production agents: fill eight hours, count evaluations completed, treat shortfall as a productivity problem. Grading is cognitive inspection work, not ticket handling. Throughput ceilings are real, fatigue shifts thresholds, and a team that "keeps up" on volume can be silently destroying measurement quality.
npx rulebase-skills install cx-reviewer-workloadWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “how many QA reviewers do we need”
How it works
The method, in the order the skill runs it. The full procedure — tables, worked examples and the edge cases — is in the skill itself.
Step 1: Measure grading time, not assumptions
Before headcount math, observe: Non-grading time.
Step 2: Set throughput ceilings, not targets
A ceiling is the maximum sustainable count before quality checks fail. A target is what management wants. When target exceeds ceiling, quality loses.
Step 3: Schedule calibration as production work
Calibration is not training that happens once. It is ongoing measurement of the measurement instrument.
Step 4: Detect fatigue and drift over a shift
Fatigue rarely announces itself as "I am tired". It shows up in the data.
Step 5: Do not mirror production staffing ratios
Production staffing answers "how many contacts arrive". QA staffing answers "how many inspections can we perform at defined quality".
Step 6: Quality-of-grading checks alongside volume
Every capacity report should pair: Volume, Quality.
Related skills
Free and open source, and vendor-neutral — it reads the conversations from whichever helpdesk you already run. Browse all 149 skills · connect your helpdesk over MCP · source on GitHub
