Acceptance Sampling is a statistical inspection approach used to decide whether a lot should be accepted or rejected based on a sample rather than inspecting every unit.
It can be useful when:
- 100% inspection is expensive;
- testing is destructive;
- inspection time is limited;
- supplier performance is being monitored.
Acceptance Sampling does not improve the process by itself.
It is a decision method for lots.
Define the lot and sample plan
A sampling plan should define:
- lot size;
- sample size;
- selection method;
- acceptance number;
- rejection number.
For an attribute plan, the team may count defective units or defects in the sample.
The plan should be established before seeing the sample result.
Select samples randomly
The sample should represent the lot.
Avoid selecting only:
- easy-to-reach units;
- top layers;
- first pieces;
- known good material.
Biased sampling reduces the value of the decision.
The selection method should give units a reasonable chance of being included.
Understand producer and consumer risk
Sampling creates uncertainty.
Two important risks are:
- producer risk: rejecting a lot that is actually acceptable;
- consumer risk: accepting a lot that is actually poor.
The sampling plan balances these risks.
A smaller sample is usually cheaper but provides less information.
Understand AQL carefully
Many acceptance-sampling systems use an Acceptable Quality Limit, or AQL, as part of plan selection.
AQL should not be interpreted as a statement that defects are desirable or that the process may freely produce that percentage defective.
It is a parameter used in the statistical design of the sampling plan.
The organization should still pursue capable processes and prevention.
Use the right standard and plan
Formal acceptance sampling may rely on recognized standards and tables.
Organizations should use the applicable customer, regulatory, industry, or internal standard rather than inventing sample sizes without understanding the statistical risk.
The exact plan depends on:
- lot size;
- inspection level;
- risk;
- historical performance;
- type of data.
Connect to supplier quality
Supplier Quality Management may use incoming sampling when supplier capability is uncertain or the consequence of a defect warrants additional verification.
As supplier performance improves, inspection strategy can be adjusted based on evidence.
Receiving inspection should not become a substitute for supplier process control.
Use stronger controls for critical characteristics
High-risk characteristics may require:
- process controls;
- certification;
- mistake-proofing;
- capability evidence;
- 100% automated verification.
Quality at the Source emphasizes prevention and control where the work is created rather than relying only on downstream inspection.
Sampling should match the consequence of failure.
Respond when a lot fails
A rejected lot may require:
- containment;
- supplier notification;
- sorting;
- rework;
- return;
- corrective action.
CAPA may be appropriate when the failure requires documented causal investigation and effectiveness verification.
The lot decision is the beginning of the response, not the end.
Common mistakes
Choosing convenient rather than random samples, using arbitrary sample sizes, treating AQL as a defect target, ignoring producer and consumer risk, using sampling for critical characteristics without risk review, and relying on incoming inspection instead of supplier process improvement are common mistakes.
Practical sequence
- define the lot.
- define the characteristic.
- select the applicable sampling standard or plan.
- determine sample size and acceptance criteria.
- select units randomly.
- inspect consistently.
- apply acceptance or rejection rules.
- contain rejected material.
- investigate recurring failures.
- adjust supplier and process controls based on evidence.
The practical lesson
Acceptance Sampling is a controlled decision method under uncertainty.
It should protect the customer while preserving focus on the more important goal: preventing defects in the process that creates them.