Acceptance sampling is a quality-control method that decides whether to accept or reject an entire lot of product by inspecting only a random sample of it, rather than checking every unit. It exists because 100 percent inspection is often impractical, expensive, or, in the case of destructive testing, impossible. By inspecting a statistically chosen sample and applying a clear accept-or-reject rule, acceptance sampling gives a defensible decision about lot quality at a fraction of the cost of full inspection. It is widely used at incoming goods, between process stages, and before shipment.
A single sampling plan is defined by two numbers: the sample size (n) and the acceptance number (c). You draw n units at random from the lot, inspect them, and count the defectives. If the number of defectives is less than or equal to c, you accept the whole lot; if it exceeds c, you reject it. For example, a plan of n = 80, c = 2 means: inspect 80 units, accept the lot if 2 or fewer are defective, reject it if 3 or more are. The rule is simple, but the choice of n and c is where the statistics live.
The central design parameter is the Acceptable Quality Level (AQL): the worst defect rate that is still considered acceptable as a process average, and which a good sampling plan should almost always accept. Every plan balances two opposing risks:
You cannot drive both to zero with a finite sample; tightening one loosens the other. The sampling plan is the negotiated compromise, and standards such as ANSI/ASQ Z1.4 provide pre-computed plans for chosen AQLs and lot sizes so teams do not derive them by hand.
The behavior of any sampling plan is summarized by its operating characteristic curve, which plots the probability of accepting a lot against the lot's true defect rate. An ideal plan would accept every lot up to the AQL and reject everything worse, a perfect step. Real OC curves are S-shaped: acceptance probability falls gradually as quality worsens. A larger sample size steepens the curve, moving it closer to the ideal and sharpening discrimination between good and bad lots, which is precisely what you pay for with more inspection.
Suppose a plant receives lots of 2,000 connectors and uses a plan of n = 125, c = 3. A supplier shipment actually contains 2 percent defectives. On average a sample of 125 would contain 125 x 0.02 = 2.5 defectives. Because 2.5 is below the acceptance number of 3, lots at this quality level will be accepted most of the time, though not always, since the actual count in any single sample varies around that average. Now imagine a bad lot at 6 percent defective: the expected sample count is 125 x 0.06 = 7.5, well above c = 3, so such lots are rejected most of the time. The plan discriminates: it usually passes 2 percent lots and usually stops 6 percent lots, with the OC curve describing exactly how often each outcome occurs.
Acceptance sampling is a gate, not an improvement tool. It sorts lots, but it does nothing to make the process that produced them better. Modern quality thinking treats it as a stopgap and pushes upstream toward prevention: a capable, monitored process should make incoming inspection increasingly unnecessary. That is why statistical process control and process capability matter more in the long run: they attack defects at the source rather than screening them at the door. When sampling does reject a lot, a Pareto analysis of the defects and one of the seven basic quality tools points to the cause, and a control plan keeps the fix in place.
Fabrico does not run sampling plans, compute OC curves, or perform the statistical inspection itself. Its contribution is upstream, where the real leverage is. By giving you a live, accurate picture of the process, Fabrico helps you build the process capability that makes lot-by-lot screening less and less necessary. Its real-time OEE and production monitoring captures output and quality losses continuously, including on machines with no PLC through computer-vision monitoring, and tracks scrap rate so defect trends are visible long before a lot is ever sampled. Its CMMS keeps the equipment that drives quality maintained through proactive maintenance. Sampling decides whether a lot passes; Fabrico helps make sure it deserves to.
It is cheaper and faster, and for destructive tests it is the only option, but it does not guarantee a defect-free lot. Full inspection is more thorough yet costly, error-prone at scale, and impossible when testing destroys the unit. Sampling is the practical choice when the cost of full inspection outweighs the residual risk, which the sampling plan quantifies.
The Acceptable Quality Level is the worst process-average defect rate that is still treated as acceptable, and which a well-chosen sampling plan should accept the large majority of the time. It is a design input to the plan, not a target to aim for, and it should not be read as permission to run at that defect level.
No. Acceptance sampling is a statistical inspection procedure carried out by your quality team using standards such as ANSI/ASQ Z1.4. Fabrico is a real-time OEE monitoring and CMMS platform, so its role is to improve and stabilize the process upstream, reducing how much lot screening you need in the first place.
Want to make incoming inspection less necessary by building a capable, monitored process? Book a Fabrico demo and see how real-time OEE and scrap tracking move quality upstream where it is cheapest.
Login
Blog