Defects At a generic level, this refers to the proportion of defects, or instances of non-conformance to quality specifications, in a given context. Measured as a ratio or a percentage (or at times parts per million), the numerator as well as denominator in this ratio have multiple ways of being defined.

• DPU (Defects Per Unit): Total Defects divided by the total number of units / transactions
• DU (Defective Units): Number of units with at least one defect, divided by the total number of units
• DPMO (Defects Per Million Opportunities): Total Defects divided by (Total Units times Opportunities per Unit times 1,000,000)

Ratio / Percentage

Accuracy

NPMO (non-conformance per million opportunities) often used synonymously to DPMO

Quality

Opportunities per Unit

• Failure Mode Effect Analysis – FMEA – also called “what could go wrong” reviews
• Root Cause Analysis – RCA – using techniques like Fishbone Diagrams
• Cause – Impact Matrix – which plots a many-to-many relationship between multiple causes and multiple impacts / symptoms / effects.
• Pareto analysis – which plots the top “n” causes of defects in decreasing order of importance (measured usually as the number of defects caused)
• 5 Whys – an industry practice of asking “why” at least 5 times, so as to get to an actionable root cause rather than try and solve at an interim cause (or just a symptom) and leave the eventual cause unsolved

• Choice of Defects, Defective Units and Defects per Unit:
  • Defective Units are of closest direct relevance to the Customer of this process under review – they expect a defect free unit. Therefore quality as seen by the customer, or wastage of material (with the work item being written off) is best measured with DU data.
  • Defects (or Total Defects) is best used when addressing inefficiency and wastage of labour, because each defect causes waste, regardless of whether they are clustered together in a small number of Defective Units or not.
  • Defects per Unit is a good measure of complexity of design or specifications, because a high number suggests there are “too many ways to get this wrong”.
• Choice of Discrete vs. Continuous measures:
  • Some specifications, like correctness of names and addresses, have a discrete outcome in terms of accuracy – they are either right or wrong, and “almost right” doesn't help. But others, like being on time, are more complex. In such situations, a basic measure of whether a task was completed on time is a good start, but richer insights can be obtained if one looks at actual time taken.
  • By looking at the actual delay of late cases, one can further segment and conduct special cause analysis and potentially get closer to multiple root causes of delays.
  • By looking at cases that get delivered ahead of time, one can assess if resources could be better used elsewhere, or if more competitive promises on turn around time can in fact be made to the customer.

• Customer CTQs – Defining Defect, Unit and Opportunity
• Zero Defects Vs. Six Sigma
• 3.4 PER MILLION:Perusing Process Performance Metrics
• Waste of Defects; causes, symptoms, examples and solutions
• "The Quality Veterans"
• Beyond Six Sigma – A Control Chart for Tracking Defects per Billion Opportunities (DPBO)
• The Concept of Zero Defects in Quality Management | Lean Six Sigma
• How a Cause and Effect Diagram Helped Reduce Defects By 19%
• The Quest for Zero Defects
• Prevent Recurring Defects with Root Cause Analysis
• Book: Call Centre Management on Fast Forward: Succeeding in Today's Dynamic Environment