ISO 28593:2017 – Acceptance Sampling Procedures by Attributes — Accept-Zero Sampling System Based on Credit Principle for Controlling Outgoing Quality

Understanding the Credit-Based Accept-Zero Sampling System

ISO 28593:2017 introduces a novel accept-zero (c=0) sampling system that uses a credit principle to dynamically adjust inspection intensity based on supplier quality history. Unlike traditional fixed-sample-size accept-zero plans (such as those in ISO 2859-1), the credit-based approach rewards consistent quality performance by reducing inspection frequency, while rapidly escalating inspection when nonconformities are detected. The standard is designed for situations where the supplier controls the production process and the buyer performs incoming inspection.

The credit principle works like a driver’s license points system: each conforming lot adds credits, each nonconforming lot deducts credits. When credits accumulate above certain thresholds, the inspection switches from normal to reduced levels. Falling below thresholds triggers tightened inspection.

Operating Mechanism and Switching Rules

Each supplier begins with a base credit of 10 points. For each submitted lot found conforming, 0.5 credits are added (up to a maximum of 20 credits). For each nonconforming lot, 4 credits are deducted. The inspection level switches based on the current credit score: normal level (6-14 credits), reduced level (15-20 credits), and tightened level (0-5 credits). If credits fall below zero, the standard recommends disqualification of the supplier pending corrective action.

Credit Score	Inspection Level	Sample Size Code	Switching Condition
15 – 20	Reduced	One letter lower than normal	Credit ≥ 15 for 10 consecutive lots
6 – 14	Normal	Per AQL and lot size table	Default starting level
1 – 5	Tightened	One letter higher than normal	Credit ≤ 5 after any lot
≤ 0	Disqualification	100% inspection required	Immediate suspension

The credit-based system offers significant efficiency advantages over traditional switching rules (from ISO 2859-1). Simulation studies show a 30-40% reduction in total inspection costs for consistently high-quality suppliers while maintaining the same consumer’s risk of accepting a nonconforming lot.

Statistical Properties and Operating Characteristic Curves

The standard provides OC curves for each inspection level. For normal inspection at AQL = 1.0%, the probability of accepting a lot with 1% nonconforming is approximately 0.95, while a lot with 5% nonconforming has less than 0.10 probability of acceptance. The accept-zero feature (c=0) means that any nonconforming item found in the sample leads to lot rejection, providing strong quality signals to the supplier.

The credit system is sensitive to lot size variations. For continuous production with consistent lot sizes, the credit accumulation rate is predictable. However, if lot sizes vary by more than a factor of 3, the standard recommends normalizing credit additions/deductions proportional to lot size to maintain consistent statistical discrimination.

Engineering Implementation Considerations

When implementing ISO 28593 in a quality management system, engineers should configure the inspection database to track credits per supplier per product family. The standard recommends separate credit tracking for each distinct manufacturing process, as a single supplier may have different quality levels across different production lines. Automated credit calculation integrated with the ERP system’s goods-receipt module enables real-time inspection level determination.

Do not apply the credit system to safety-critical components where zero-defect performance is mandatory regardless of credit score. For such applications, 100% inspection or process validation per ISO 9001 clause 8.5.2 is required irrespective of historical quality performance.

Operating Characteristic Curves and Statistical Discrimination

The statistical performance of the credit-based sampling system is characterized by its operating characteristic (OC) curves, which show the probability of lot acceptance as a function of incoming quality level. For the normal inspection level at AQL = 1.0%, the OC curve provides a probability of acceptance of approximately 0.95 at the AQL (producer’s risk α = 5%), dropping rapidly to approximately 0.50 at 3.0% nonconforming and approximately 0.10 at 7.0% nonconforming (consumer’s risk β = 10%). The accept-zero criterion (c=0) means that any single nonconforming item found in the sample results in lot rejection, providing strong incentives for suppliers to maintain zero-defect production. The credit system’s dynamic nature means that the effective OC curve shifts over time: for a supplier with high credit (reduced inspection), the OC curve is somewhat steeper (better discrimination) due to larger effective sample size per lot over the multi-lot sequence, while for a low-credit supplier under tightened inspection, the OC curve shifts toward higher protection levels for the consumer. The standard provides tabulated OC curves for each inspection level, enabling quality engineers to select appropriate AQL values based on the acceptable risk levels for both producer and consumer. The mathematical basis of the credit system follows a Markov chain model where the credit score evolves probabilistically based on the true process quality level, with known steady-state distributions and transition probabilities that have been validated through extensive Monte Carlo simulation studies.

Monte Carlo simulation studies of the ISO 28593 credit-based sampling system demonstrate that it achieves a 30-40% reduction in total inspection costs compared to traditional ANSI/ASQ Z1.4 (ISO 2859-1) sampling, while maintaining equivalent or better consumer protection levels. The savings are most pronounced for consistently high-quality suppliers, who may experience inspection rates as low as 20% of the traditional sample size after accumulating sufficient credit.

Implementation Requirements for Quality Management Systems

Successful implementation of ISO 28593 requires integration with the organization’s quality management system. The standard specifies minimum requirements for: documentation of the credit calculation methodology, training of inspection personnel on the switching rules, periodic audit of credit score accuracy (at least quarterly), and management review of supplier credit trends (at least annually). For each supplier, the quality department must maintain a credit ledger that includes: supplier identification, product family, date of each lot receipt, lot size and sample size, number of nonconforming items found (zero for conforming lots), disposition decision, credit adjustment, and resulting credit score. The standard recommends implementing automated credit tracking within the ERP system’s supplier evaluation module, with real-time alerts when a supplier enters tightened inspection status requiring mandatory corrective action per ISO 9001 clause 10.2. The quality system must also define escalation procedures for suppliers whose credit score reaches zero, typically requiring a formal quality improvement plan with defined milestones and 100% inspection at the supplier’s expense until credit is restored above the disqualification threshold.

One common pitfall in implementing the credit-based system is failing to reset credit scores when a supplier changes a critical process element (new production line, new raw material source, or major equipment change). The standard requires that credit be reset to the base level of 10 when the supplier’s production process undergoes a significant change that could affect product quality, as historical quality data is no longer predictive of future performance under the changed conditions.

FAQ

Q: How does ISO 28593 differ from the c=0 plans in ANSI/ASQ Q3?
A: ISO 28593 uses a dynamic credit system that adjusts over time, while ANSI/ASQ Q3 uses fixed sample sizes for given lot sizes. The credit system provides more flexibility for suppliers with proven quality track records.

Q: Can the credit system be gamed by suppliers?
A: The standard includes safeguards against gaming. The 4-credit penalty for a nonconforming lot far exceeds the 0.5-credit reward for a conforming lot, making it economically irrational to submit nonconforming lots.

Q: What AQL values are covered?
A: The standard covers AQL values from 0.01% to 10% in the standard series per ISO 2859-1.