Cost of Quality (COQ): The Juran-Crosby Framework Applied to Software
Updated May 2026. Sources: Juran (1951), Feigenbaum (1956), Crosby (1979), ASQ current guidance.
The framework in one paragraph
Cost of Quality (COQ) is the total cost an organisation incurs because of quality, regardless of which side of the ledger the cost falls on. COQ splits into Cost of Good Quality (COGQ = prevention plus appraisal investment) and Cost of Poor Quality (COPQ = internal failure plus external failure spending). The PAF model is the operational decomposition. Crosby's central thesis is that COGQ investment is self-funding through COPQ reduction; mature operations show small COPQ and a healthy COGQ, total COQ around 4 to 10% of revenue.
The five-decade history
Joseph Juran first defined the four-bucket cost structure in the 1951 first edition of his Quality Control Handbook. The handbook proposed organising the cost of quality across prevention activities (training, planning, process design), appraisal activities (inspection, testing, audit), and the two categories of failure cost. Juran did not yet use the modern terminology; he called the categories "cost of conformance" and "cost of non-conformance" in early editions.
Armand Feigenbaum coined the specific phrase "cost of quality" in his 1956 Harvard Business Review article Total Quality Control, expanded into the book of the same name in 1961. Feigenbaum's contribution was to argue that quality is a cross-functional responsibility, not just a manufacturing-floor inspection problem; the COQ framework provided the financial discipline to make that argument concretely.
Philip Crosby's Quality Is Free (1979) popularised the PAF acronym and made the central thesis explicit: total COQ stays roughly constant across operating points, so moving spend from failure to prevention does not increase total cost. The argument resonated with senior management in a way that Juran's and Feigenbaum's more technical formulations had not. Crosby's book sold over 2 million copies and remains the most-cited single source on COQ economics.
The vocabulary that matters
| Term | What it means | Typical share (mature) |
|---|---|---|
| COQ | Total Cost of Quality (everything) | 4 to 10% of revenue |
| COGQ | Cost of Good Quality (prevention + appraisal) | 2 to 6% of revenue |
| COPQ | Cost of Poor Quality (internal + external failure) | 2 to 4% of revenue |
| PAF | Prevention, Appraisal, Failure (the four-bucket decomposition) | see below |
| Prevention | Spend on stopping defects before they exist | 4 to 6% of payroll |
| Appraisal | Spend on confirming quality | 6 to 10% of payroll |
| Internal failure | Defects caught before customer sees them | 4 to 8% of payroll |
| External failure | Defects discovered by customers | 1 to 3% of payroll |
Crosby's thesis: quality is free
The central claim in Quality Is Free is that COGQ investment is self-funding through COPQ reduction. A dollar invested in prevention typically saves multiple dollars in failure cost; total COQ either stays flat or drops. Crosby illustrated the claim with case studies from ITT operations he had led as Director of Quality. The case studies showed COQ improvements from 20% of revenue to 5 to 8% over 24 to 36 month programmes, with prevention spend roughly tripling while failure spend dropped by 75% or more.
Modern measurement broadly supports the thesis, with one important caveat. The total-COQ-constant claim holds well for operations far from the optimum (i.e. starting from a high-COPQ baseline), where prevention investment really does fund itself through failure reduction. For operations already near the optimum, further prevention investment shows diminishing returns; the curve flattens and total COQ becomes harder to reduce. The implication is that the Crosby thesis is most strongly true at the operations that need it most.
The software application
Software engineering teams that have adopted the COQ framework (typically large enterprise SaaS with internal quality engineering functions) report findings consistent with the cross-industry pattern. The PAF category line items map cleanly to software-specific spend; the typical mature-operation distribution is similar to manufacturing benchmarks; the Crosby self-funding thesis holds at the level of moving from immature to mature.
The two software-specific adaptations worth noting. First, software has a higher latent appraisal cost (the ongoing maintenance of automated tests is not analogous to a manufacturing inspection line that runs and stops). The mature software operation typically lands at slightly higher appraisal spend than the manufacturing equivalent. Second, software has a lower latent external-failure cost (a software defect rarely sends a customer to hospital). The mature software operation typically lands at slightly lower external-failure cost than the manufacturing equivalent.
The practical implication for software-engineering leaders adopting COQ: benchmark against software-industry data rather than against the ASQ manufacturing curves. DORA State of DevOps is the most reliable software-industry benchmark for the failure-rate side; Capers Jones' Namcook publications are the most reliable for the broader PAF distribution. The COPQ calculator page and the PAF model page cover the operational details of running the measurement for the first time.
Sources
- Juran, J. M. Quality Control Handbook. McGraw-Hill, 1951 (and subsequent editions through 1999 as Juran's Quality Handbook).
- Feigenbaum, A. Total Quality Control. Harvard Business Review, November 1956; book edition McGraw-Hill, 1961.
- Crosby, P. B. Quality Is Free. McGraw-Hill, 1979.
- American Society for Quality. Cost of Quality (COQ) Guidance. asq.org/quality-resources/cost-of-quality.
- Google DORA. State of DevOps Report 2024.
- Jones, C. Applied Software Measurement. 3rd ed. McGraw-Hill, 2008.
Frequently asked questions
What is Cost of Quality?▼
The total cost an organisation incurs because of quality. COQ splits into COGQ (prevention + appraisal investment) and COPQ (internal + external failure spending). The PAF model is the operational decomposition.
Who invented Cost of Quality?▼
Joseph Juran defined the four-bucket cost structure in 1951. Armand Feigenbaum coined the phrase 'cost of quality' in 1956 HBR. Philip Crosby popularised the PAF acronym and the quality-is-free thesis in 1979.
What is a typical COQ percentage for a mature operation?▼
ASQ places mature operations at 4 to 10% of revenue, with the best below 4%. Immature operations run at 15 to 25%. The improvement trajectory typically takes 18 to 36 months.
How does COQ apply to software engineering?▼
Direct mappings. Prevention: design review, training, threat modelling, accessibility audits, tech-debt paydown. Appraisal: test maintenance, code review, QA, security scanning, observability. Internal failure: pre-release rework. External failure: incidents, hotfixes, support, churn, penalties.
Is COQ the same as COPQ?▼
No. COPQ is the failure side (internal + external failure). COGQ is the investment side (prevention + appraisal). COQ is the total. Separating them is the point: COGQ is investment, COPQ is loss, the goal is to substitute the first for the second.
Does Crosby's quality-is-free thesis hold in software?▼
Broadly yes, with the caveat that the self-funding claim is strongest at operations starting from a high-COPQ baseline. Software teams moving from immature to mature typically see the predicted COGQ-funded COPQ reduction. Operations near the optimum see diminishing returns from further prevention investment.