Process Inputs
Example Data Table
| Category | Defects | Notes |
|---|---|---|
| Surface scratch | 48 | Often after packing |
| Mislabel | 21 | Shift B spike |
| Loose fastener | 17 | Torque checks inconsistent |
| Color mismatch | 9 | Supplier mix-up |
| Seal leak | 5 | Tool wear suspected |
Formula Used
- Defects per 1,000 = (Total Defects ÷ Units Inspected) × 1,000
- DPMO = (Total Defects ÷ Units Inspected) × 1,000,000 (assumes one opportunity per unit)
- Estimated Sigma = NORMSINV(1 − DPMO/1,000,000) + 1.5 shift
- Category Share % = (Category Defects ÷ Total Defects) × 100
- Cumulative % = running total of shares in descending order
- Suspect Score = Likelihood × Impact × (1 + Evidence/5)
How to Use This Calculator
- Enter the process, period, and a clear problem statement.
- Add inspected units and your defect totals if known.
- Fill defect categories so Pareto highlights dominant drivers.
- Score the 6M factors based on what you observe.
- Add specific suspects and score them with evidence quality.
- Click Find Causes to see ranked outputs above.
- Export CSV for sharing, or PDF for reporting.
Professional Insights
Defect signals and process stability
Quality issues rarely come from one event; they build when variation grows and controls drift. Enter inspected units and defects to quantify burden and compare periods consistently. Defects per 1,000 and DPMO convert counts into rates that stay meaningful when volume changes. The sigma estimate offers a shared stability signal, showing whether a spike is noise or a persistent shift needing action.
Pareto focus for faster containment
A category list enables a Pareto view that separates the vital few from the trivial many. Sorting categories by frequency and tracking cumulative share shows which issues dominate customer impact and rework cost. When one category reaches the 80% threshold quickly, begin containment: isolate affected lots, add temporary checks, and protect downstream steps. This focus prevents effort from scattering across minor defects.
6M scoring to structure hypotheses
Root-cause work accelerates when hypotheses are organized. The 6M framework groups potential causes into people, equipment, methods, materials, measurement, and environment. Scoring likelihood and impact promotes realistic thinking, while the evidence factor rewards observations, records, and test data. A high score is not proof; it highlights where diagnostics are most likely to pay off. Adding specific suspects turns themes into checks.
Evidence-driven verification methods
Verification should reduce uncertainty quickly and cheaply. Start with data integrity: confirm defect definitions, sampling rules, and inspection repeatability. Then run tight experiments that change one factor, such as adjusting torque, swapping a material lot, or validating gauge calibration. Record the before-and-after defect rate and capture evidence like photos, logs, and notes. Strong evidence reduces debate and speeds decisions.
Turning results into corrective actions
Use ranked outputs to build a short corrective action plan. For each top driver, assign an owner, define a test, and set a target reduction in defects per 1,000. If sigma improves, standardize settings through work instructions and training. If results stall, revisit suspects and raise evidence quality. CSV exports support reviews, while PDF snapshots preserve an audit trail for learning. Track actions weekly, and refresh the Pareto after each change. When a new category rises, repeat scoring to keep attention aligned with the latest process behavior.
FAQs
What does “vital few” mean in the results?
It lists the smallest set of defect categories that accounts for roughly 80% of total defects. Start containment and investigation there to get the fastest reduction.
Do I need to enter total defects if I already filled categories?
No. If category counts are provided and total defects is blank or zero, the calculator sums categories and uses that value for rates and Pareto.
How is the sigma level calculated here?
Sigma is estimated from DPMO using an inverse normal yield calculation, then a 1.5 sigma shift is added. It is a benchmarking indicator, not a certification result.
How should I rate evidence in the 6M and suspect scores?
Use 0–5 based on proof quality: 0 is a guess, 3 includes records or repeatable observations, and 5 includes test data or controlled experiment results.
What if two suspects have similar scores?
Increase evidence before acting: collect logs, photos, measurements, or run a small trial that isolates one factor. Then rescore and prioritize the clearer signal.
How often should I rerun the calculator?
Rerun after each significant change, and at least weekly during active issues. Use the same period definition so rates remain comparable and improvement is visible.