top of page
Working Checklist
Quality Management Checklist
A practical 50-point assessment for your quality program. Walk through each item with your team and mark where you stand — doing well, improving, or not yet.
To save or print: Press Ctrl+P (Windows) or Cmd+P (Mac), then choose "Save as PDF" or your printer.
1. Strategy & Governance
- Quality is defined as a measurable business outcome, not just a function or team.
- Quality goals are tied to specific customer or business metrics (e.g., retention, MTTR, defect escape rate).
- Leadership reviews quality KPIs at least monthly and acts on regressions.
- A clear quality policy exists and is reviewed at least annually.
- Risk-based prioritization is documented for which work gets which level of quality scrutiny.
- A single accountable owner for quality at the program level (not just the QA team).
2. Process & Documentation
- Written definition of "ready" and "done" for work items, shared across teams.
- Major processes (release, incident, change) have current SOPs that match how work actually happens.
- Process documentation is searchable, versioned, and updated when reality changes.
- Acceptance criteria are written before development starts, not after.
- Onboarding for a new engineer to ship-ready takes a documented, predictable amount of time.
- Decisions are recorded (ADRs, decision logs) so context survives turnover.
3. Quality Engineering
- Automated tests run on every commit; a red build blocks merge.
- Test pyramid is healthy: most tests are fast unit tests, fewer integration, fewer end-to-end.
- Test data is reproducible and does not depend on a fragile shared environment.
- Quality gates in CI/CD enforce coverage, security, and performance thresholds.
- Code review checks for quality concerns (testability, observability) — not just correctness.
- Feature flags allow shipping incomplete work safely behind a switch.
- Production has observability adequate to detect, diagnose, and recover from new failure modes.
- Rollback for any change is achievable within a documented timeframe.
4. Measurement & Metrics
- Defect escape rate (bugs found by users vs caught pre-release) is measured per release.
- Mean time to detect and mean time to recover (MTTD / MTTR) are tracked for production incidents.
- Quality metrics are reported alongside delivery metrics (not in isolation).
- A small number of leading indicators are watched (e.g., test flake rate, change failure rate).
- Metrics are reviewed for what they incentivize, not just what they measure.
- Targets are anchored to customer impact, not vanity numbers.
5. People & Culture
- Engineers feel safe raising quality concerns without career penalty.
- Postmortems are blameless and lead to system changes, not just individual training.
- Skill gaps in testing, observability, and review are surfaced and funded.
- New hires are paired with experienced engineers during their first months.
- Quality work is recognized in promotions and performance reviews, not just feature delivery.
- Cross-team collaboration on quality is structured (guilds, communities of practice, office hours).
6. Continuous Improvement
- A retrospective cadence exists and produces durable actions, not just feelings.
- Action items from retros are tracked to completion with an owner and a deadline.
- Improvement work has explicit budget within each sprint or planning cycle.
- Process changes are piloted on small surface area before rolling out broadly.
- External benchmarking (DORA metrics, industry standards) informs internal targets.
- A clear escalation path exists when a fix is repeatedly deferred for "later."
7. AI & Tooling
- AI tools used in development have a documented policy: when they help, when they need extra review.
- AI-generated code is reviewed with the same rigor as human-written code.
- Test generation tools are evaluated for false-confidence (passing tests that test nothing).
- Where AI is part of the product, model behavior has its own evaluation suite.
- Sensitive data flows to AI tools are governed and logged.
- Tool sprawl is periodically audited; redundant or unused tools are removed.
- Build, test, and CI tooling is reliable enough that engineers trust the signal.
How to use this checklist
Walk through it with your team once a quarter. For each item:
- ✓ Doing well — keep it, watch for drift.
- → Improving — assign an owner and a target review date.
- ✗ Not yet — decide if it matters; if it does, add it to next quarter's plan.
Product
bottom of page