
Quality Control
This framework was generated using the Framework Builder methodology. It demonstrates the five-layer DNA architecture applied to quality control for creative and technical work.
Built with the Framework Builder methodology
Download markdownThis framework was generated using the Framework Builder methodology. It demonstrates the five-layer DNA architecture applied to quality control for creative and technical work.
Layer 1: Principles Foundation
Principle 1: Quality belongs to the person making the work. The quality function verifies the system. It does not compensate for people who throw work over the wall.
Principle 2: Nothing ships without a second set of eyes. Review is not optional, whatever the seniority or deadline.
Principle 3: Ask which process allowed it, not who did it. Blame teaches people to hide problems, and hidden problems turn into serious ones.
Principle 4: "Good" is written down before work starts. If the team cannot describe it, they are guessing, and quality becomes a matter of opinion.
Principle 5: Prevention costs less than rework, and rework costs less than losing customers. Spend the money early.
Layer 2: Systematic Approach
Step 1: Identify the critical control points. Where can a defect be introduced or caught? A typical set: requirements sign-off, peer review, an integration or staging check and a pre-launch checklist.
Step 2: Route by size of work. Small maintenance tasks take a lightweight path with a single reviewer. New builds take the full path with structured review, staging verification and a client preview.
Step 3: Write pass/fail criteria for each control point. Put them in templates so reviewers are not guessing what to check.
Step 4: Write a definition of done per work type. Makers check their own work against it before asking for review.
Step 5: Investigate each escape. When a defect reaches a customer, hold a short root-cause review, trace it to the process gap and update the checklist or template.
Layer 3: Force Multipliers
Multiplier 1: Automate the checkable. Linting and tests on every submission catch formatting errors and basic logic bugs before a human spends time on them. This is error-proofing.
Multiplier 2: Definition-of-done documents. One per project type, short enough to use.
Multiplier 3: A fifteen-minute root-cause review for every production defect. The output is a process change, not a name.
Multiplier 4: Fix the process, not the inspection. Before adding a review step, ask what created the defect. A fixed process needs fewer inspections.
Multiplier 5: Two lanes, not one. Keep the lightweight path genuinely light. If every maintenance task needs the full path, people will route around the system. Reserve structured review and client preview for new builds.
Layer 4: Success Metrics
Positive indicators:
- Production defects per project in the first 30 days stay under your target (for example, under three)
- Review rejection rate settles at a level that shows clear standards
- Average rework hours per project fall
- Client-reported issues per quarter fall
Failure signals:
- Rejection rates are high because standards are unclear
- Defects fall while rework rises (prevention is working but standards may be too rigid)
- Reviews are skipped under deadline
- Every defect ends with a person being named
- New inspection steps keep appearing without a change in defect type
Make each phase small enough to finish. A team that feels overwhelmed abandons the framework. Each phase should leave something working that the next one builds on.
Layer 5: Implementation Guidance
Roll out in three phases over about 90 days. Weeks 1 to 2: write definitions of done for the three most common project types and automate the checks. Weeks 3 to 6: formalize the review template and start tracking the four metrics. Weeks 7 to 12: begin root-cause reviews and hold the first monthly quality review against baseline data.
Review trends every six months. Read all four metrics together, not one at a time.
Common mistake: treating quality as overhead. Calculate what rework and lost clients cost before cutting prevention.
Edge case: a deadline collides with review. Shorten the scope of work, not the review. If the review must be reduced, record the exception and the owner of the risk.
Edge case: a senior person resists review. Apply the rule to everyone, and let the checklist, not the reviewer, carry the standard.
Common mistake: quality as someone else's job. When a separate team catches the problems after the work is done, makers stop thinking about quality. Keep ownership with the people creating the work.
Common mistake: defining quality subjectively. "I'll know it when I see it" is a guess. Every work type needs a written definition of done with measurable criteria, or quality varies by person and mood.
Common mistake: adding inspection instead of fixing the process. Another review step catches the same defects in a different place and adds cost and time. Change what produces the defect and you need fewer checks.
Common mistake: holding individuals accountable for systemic failures. If the process allowed the mistake, the process is the problem. Blame creates fear, and fear hides problems until they become catastrophic.
Built with the Framework Builder methodology. Get the skill →