Proportion nonconforming is the number of inspected units that fail at least one specified requirement divided by the total number of units inspected, such as orders checked before release to production.
In an order-verification process, each order counts once as conforming or nonconforming. An order containing both an incorrect material and an incorrect quantity contributes one nonconforming order, even though it contains two defects. The measurement plan must define the inspection unit and the criteria for accepting it.
For example, suppose a pilot checks 200 orders and finds 10 orders containing a total of 14 discrepancies. The proportion nonconforming is 10 ÷ 200 = 0.05, or 5%. The same records yield 14 ÷ 200 = 0.07 defects per order. These measures answer different operational questions: how often orders fail acceptance and how many discrepancies occur across the inspected workload.
Within DMAIC, proportion nonconforming can provide a baseline during Measure and track performance after a corrective action. Comparisons require consistent inspection criteria and a sampling approach that represents the orders under review. Inspecting only disputed orders would produce a biased estimate of overall performance.
An operational definition establishes the counting rules. A Pareto chart separates discrepancy categories for investigation, while a p-chart monitors the proportion nonconforming over time, with control limits that account for subgroup sample size.
