Skills Quality assurance
QA programme maturity
High review volume is the most common false signal of maturity. Teams that score 10,000 conversations a month but never calibrate, never action auto-fails, and cannot link scores to CSAT or repeat contact are not mature — they are busy.
npx rulebase-skills install cx-qa-program-maturityWhen to use it
Reach for this when someone says any of these — they are the phrases the skill itself triggers on:
- “how mature is our QA programme”
- “QA maturity assessment”
- “are we ready to use QA for bonuses”
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: Name the decisions scores actually drive
Ask explicitly — do not accept the mission statement.
Step 2: Score five dimensions
Rate each 1–4 (1 = ad hoc, 4 = reliable and evidenced). No dimension above 2 if gating decisions run on that layer without upgrade.
Step 3: Overall maturity — limited by weakest load-bearing dimension
Examples: Coaching programme, Bonus gating, Compliance.
Step 5: Roadmap by gap — not by ambition
Prioritise fixes that unlock the next decision the business wants to make safely.
Step 6: Assess AI grading maturity separately
If AI grades production traffic, add: Model/version pinned on evaluations, Human benchmark sample with agreement tracked, Segment differential agreement (especially voice), Contested evaluation workflow includes AI-specific remand reasons.
Step 7: Report maturity honestly
Present to leadership: Load-bearing dimensions for each claimed use of QA, Explicit blockers, 90-day priorities.
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
