Zero Defect (c=0) Sampling Plan — A Sampling Method That Rejects All Nonconforming Products

Abstract: Conventional AQL sampling plans allow for a certain number of nonconformities within a batch (Ac≥1), which inherently conflicts with the "zero defect" quality philosophy in industries such as automotive and medical devices. The c=0 zero defect sampling plan sets the acceptance number to 0 — if a single nonconforming product is found during sampling, the entire batch is rejected. This approach ensures that no nonconforming products are released from a batch deemed acceptable. This article clarifies the differences between c=0 and AQL plans, explains the logic for determining sample size, and provides a four-step implementation method. It also demonstrates the complete decision-making process using an example of an automotive component batch.

1. Limitations of AQL Plans: Accepting Nonconformities in Qualified Batches

Conventional attribute sampling (GB/T 2828.1 / ISO 2859) is centered around AQL, with the sampling plan determined by three elements: (sample size n, acceptance number Ac, rejection number Re), where the acceptance number Ac is usually greater than or equal to 1. For example, in the plan (125, 3, 4): 125 items are randomly selected from the batch, and if the number of nonconforming products does not exceed 3, the entire batch is deemed acceptable — meaning that a batch of products can "legally" contain 3 nonconforming items.

For general industrial products, this aligns with the principle of quality economics; however, for safety components, medical consumables, and braking parts, customers cannot accept "nonconformities in a qualified batch." When customers explicitly require zero defects or when historical complaints are concentrated on certain characteristics, the conventional AQL plan is insufficient, and a stricter c=0 plan should be adopted.

2. What is the c=0 Plan: Zero Acceptance Number Judging Logic

The core rule of the c=0 (Zero Acceptance Number) plan is simple: the acceptance number Ac=0, and the rejection number Re=1. If a single nonconforming product is found during sampling, the entire batch is judged as nonconforming and is subject to return, full inspection, or sorting.

Compared to conventional AQL plans, the c=0 plan has three key differences. First, the judging rules are different: conventional plans allow "a small number of nonconformities" (Ac≥1), while the c=0 plan allows "none at all." Second, the sample sizes differ: under the same AQL, the c=0 plan typically requires a larger sample size than the corresponding GB/T 2828.1 plan — because with an acceptance number of 0, more samples are needed to achieve the same batch quality discrimination. Third, the application scenarios differ: conventional plans are suitable for general materials, while the c=0 plan is suitable for safety characteristics, key characteristics, and materials explicitly designated as zero defect by the customer.

It is important to clarify that the c=0 plan is not a full inspection; it remains a sampling inspection. Its role is to use a larger sample size and stricter judging rules to minimize the probability of releasing a nonconforming batch, not to reduce the probability of nonconformities to zero.

3. How to Determine Sample Size: Use Tables, Not Guesswork

The sample size for the c=0 plan is determined by the batch size and AQL, and the industry standard is the zero defect sampling table (such as the c=0 plan table in QS-9000 Appendix C). The table lookup logic is similar to GB/T 2828.1: first, find the batch size interval, then read the sample size n from the corresponding AQL column, with the judging rule fixed as (n, 0, 1).

There are three key points to consider when looking up the table. First, the AQL is determined by the product characteristics: safety and regulatory characteristics should be set at 0.65 or lower, while general appearance characteristics can be set at 1.0 to 2.5. Second, the batch size interval and sample size are not proportional: if the batch size increases from 3201 to 10000, the sample size may remain unchanged. Do not rely on the intuition of "sampling proportionally" to increase the sample size. Third, the c=0 plan typically does not have the conventional switching rules for relaxed or tightened inspection, but instead uses "cumulative qualified batches" to dynamically adjust the inspection frequency — for example, after 10 consecutive batches with zero nonconformities, the inspection frequency can be adjusted from sampling each batch to sampling every 3 batches, but the sample size for each inspection should still follow the standard table.

4. Four-Step Implementation Method

Step 1: Identify Applicable Objects. List the materials and characteristics that require zero defect control — prioritize safety components, customer-designated components, and historically high-frequency nonconforming components. Continue to use the conventional AQL plan for general materials to avoid uncontrolled inspection costs.

Step 2: Determine the AQL Value. Set the AQL value based on customer requirements, product risk levels, and historical quality performance. The automotive industry commonly uses 0.65 or 1.0, and for characteristics involving regulatory safety, it is recommended to use 0.4 or lower.

Step 3: Determine the Sample Size from the Table. Based on the batch size interval and AQL value, read the sample size n from the c=0 sampling table. For example, if the batch size is 5000 and AQL=0.65, the table lookup yields n=200, resulting in the plan (200, 0, 1).

Step 4: Execute, Judge, and Handle. Randomly select 200 items for inspection: if the number of nonconforming products is 0, the batch is accepted; if the number of nonconforming products is ≥1, the entire batch is rejected, triggering the nonconforming product handling process (return or full inspection and sorting), and the supplier's corrective actions are escalated. Record the results for each batch, and after a cumulative number of consecutive qualified batches meet the criteria, adjust the inspection frequency according to the procedure.

5. An Example: Incoming Quality Control for Automotive Steering Rods

A certain automotive parts factory implements a c=0 plan for steering rods (safety components). The batch size is 5000, and the customer requires AQL=0.65. According to the c=0 sampling table, for a batch size of 3201 to 10000 and AQL=0.65, the corresponding sample size is 200, resulting in the plan (200, 0, 1).

IQC randomly selects 200 items for inspection and finds 1 item with a thread size out of tolerance — according to the c=0 rule, the entire batch is judged as nonconforming. The handling process: the batch is returned to the supplier for full inspection and sorting, and an 8D report is issued simultaneously to require corrective actions. A traceability process is also initiated for the inventory already in use. After 8 consecutive batches with zero nonconformities, the inspection frequency is adjusted from sampling each batch to sampling every 3 batches (the sample size for each inspection still follows the table).

In contrast, if the conventional AQL plan were used, 1 nonconforming item would be allowed in a sample of 200 items (e.g., plan (200, 1, 2)), and this batch would be "legally" released, with the nonconforming item directly flowing to the assembly line — this is precisely the value of the c=0 plan.

6. Two Reminders for Implementing the c=0 Plan

First, do not switch all materials to the c=0 plan. The cost of zero defects is increased inspection costs, so it should only be used for high-risk materials, those with clear customer requirements, and those with a history of nonconformities. Otherwise, the benefits do not justify the costs. Second, the c=0 plan is not about "the more you sample, the safer it is" — the sample size is determined by the standard table. Randomly increasing the sample size can undermine the statistical properties of the plan. The real defense is to drive process improvements at the supplier, ensuring that the process capability is sufficiently good — sampling is merely a verification method, and the root of zero defects lies in the process, not in the inspection.

Not letting a single nonconforming product pass is the true meaning of zero defect sampling — the c=0 plan is strict not in the inspection but in the supplier's process.

Knowledge code: 6.2.2

Version: v20260808

Author: Quality Think Tank The Quality Think Tank is dedicated to providing systematic professional knowledge, methodologies, and practical tools for quality management practitioners, helping companies continuously improve their quality capabilities.