How Do You Enforce AQL 2.5 on a Reorder Run?

Almost every quotation in this industry offers inspection at AQL 2.5. Very few say what that means on a particular lot, which is why the offer is worth little. An acceptance quality limit is not a promise about quality; it is a parameter of a sampling plan, and the plan is what does the work.

On a reorder the plan has to do a second job as well. It has to find units that differ from their lot, which is what sampling is for, and it has to show whether the lot differs from the approved run, which sampling alone cannot do. That second job is the comparison, and it is what most AQL 2.5 inspections leave out.

The confusion is understandable because the acronym is used loosely across the industry. A buyer asking for AQL 2.5 and a supplier quoting AQL 2.5 are often agreeing on one number and disagreeing on everything that gives that number meaning.

What Does AQL 2.5 Actually Say?

It names one parameter of a sampling scheme indexed by acceptance quality limit: the quality level at which the plan is designed to accept most lots. It says nothing on its own about how many units are pulled, what counts as a defect, or what happens when the lot fails.

Those three come from elsewhere: the sample size comes from the lot size and the inspection level, the defect definition comes from a written classification, and the consequence comes from the contract. An offer of AQL 2.5 that specifies none of them is an offer of a number.

There is a second reason, specific to a repeat: consistency of judgement. The same inspector, using the same plan, on the same style, over several runs, produces a series that can be read. Change the plan and the series is broken at the point of change, which is usually the point somebody most wants to read.

Why Does a Reorder Need the Same Plan as Run One?

Because the plan is what makes two runs comparable. If run one was inspected with one sample size and run five with another, the two inspection results cannot be compared, and the programme has lost the only evidence that would show a change in quality.

It also keeps the buyer honest about what a pass means. A smaller sample on a smaller lot is not a stricter inspection; it is the same acceptance quality limit judged on less evidence, and it is worth saying so rather than assuming it means more.

Level II is the general default and the one most programmes use. Level I gives a smaller sample and is used where the cost of inspection is high relative to the risk; Level III gives a larger one. Whichever is chosen, it should be chosen deliberately and written down rather than inherited.

How Is the Sample Size Derived?

From the lot size and the general inspection level, read from the standard tables. Under ISO 2859-1 at General Inspection Level II and AQL 2.5, a lot of 151 to 280 units takes a sample of 32 with an accept number of 2 and a reject number of 3, and a lot of 281 to 500 takes 50 with 3 and 4.

An inspection report that quotes those figures is reading the standard. One that quotes a round number, or a percentage, or the same sample size for every lot, is not - and the difference is visible in one line of the report.

The classification also has to be signed off by both sides before the run, not produced with the report. A classification written after the units were graded is a description of what the inspector decided, and it will be read as an attempt to justify the result rather than to explain it.

What Is a Defect Classification and Why Write It Down?

It is the written list of what counts as critical, major and minor on this product, with examples. Without it, two inspectors grade the same bag differently, and a pass on one run is not comparable with a pass on the next.

Critical

Makes the product unsafe or unlawful, or fails a regulated requirement

Major

Affects function, fit or saleability, or is visible to the customer

Minor

A workmanship deviation that a customer would not act on

Test

Would the customer return it? Yes is major; unsafe is critical

Record

Every class needs named examples on the product, not a description

Consequence

Critical: reject. Major: against the plan. Minor: against the plan

The classification has to name examples on the actual product. A general description - poor workmanship - is not a classification, because it is satisfied by any judgement and constrained by none.

There is an ordering point that saves time: do the measurement pass first on five units before grading the whole sample. Where those five show a movement, the grading pass is going to find the same story, and the inspection can be stopped and the drift question raised straight away.

How Is the Inspection Applied to a Comparison?

In two passes over the same sample. The first pass grades each unit against the defect classification and counts defects. The second pass takes the measurements and the readings against the retained reference and the tolerance table, on the comparison sheet.

Two passes, one sample. Running them on different units is the common shortcut and it weakens both: the defect count no longer describes the units that were measured, and the measurements no longer describe the units that were graded.

What Does a Reorder Inspection Add to a Standard One?

Three things: the retained reference on the table, the tolerance table with its methods, and a column for the approved-run value next to each reading. Those three turn an inspection into a comparison, and they are the whole difference.

Switching rules deserve a mention because they are part of the standard and are rarely used. Normal, tightened and reduced inspection move the stringency according to recent history, and a programme that has had two lots rejected should be running tightened inspection whether or not anyone asks for it.

How Are Defects Counted Against the Accept Number?

Defective units are counted, not defects. A unit with three minor deviations is one defective unit. Critical defects are normally treated separately: a single critical usually rejects the lot regardless of the count, because the classification says the product is unsafe or unlawful rather than merely imperfect.

Count each class separately against its own plan where the programme uses different limits per class. Where one limit is used for everything, say so, because the report will otherwise be read differently by two people.

There is a sixth question that belongs on a reorder and not on a first order: does this failure change what the next run should be? Where the answer is yes, the correction belongs in the spec file and the register rather than only in the cartons, and that is the step most often skipped after a successful rework.

What Happens When the Lot Is Rejected?

Four outcomes, in ascending order of cost: sort and rework the lot, accept it with a written exception for one run, re-inspect after rework, or reject and replace. Which one applies should be written in the contract rather than decided while the goods are waiting.

On a reorder there is a fifth question that matters more than the remedy: is the failure a defect or a drift? If the readings show the lot has moved rather than that some units are bad, sorting will not fix it, and the correction belongs in the process rather than in the cartons.

Read the photographs too, where the report carries them. A photograph is supporting evidence rather than a measurement, but it is often the fastest way to understand a defect description written by someone whose first language is not the reader's.

How Is the Inspection Report Read?

Four lines first: the sample size and where it came from, the defect count by class, the accept and reject numbers used, and the result. Then the readings. A report that gives a result without those four lines cannot be checked, which in practice means it cannot be trusted.

The standard cited matters too. A report that names ISO 2859-1, the inspection level and the acceptance quality limit is a report that can be reproduced by someone else. One that says AQL 2.5 without any of the three cannot be reproduced, and reproducing it is exactly what a dispute requires.

One more thing makes the difference in practice: who pays for a re-inspection. Deciding that in advance removes the hesitation about calling a lot failed, which is the moment when most enforcement quietly collapses.

It is also worth stating what happens when a lot passes and a customer still complains. A sampling plan accepts lots with a known level of defects, so a pass is not a warranty. Saying that once, in writing, prevents the conversation in which a passed lot is treated as a broken promise.

What Makes AQL 2.5 Enforceable Rather Than Decorative?

Three things in the contract: the standard and the level, the classification with its examples, and the consequence of a failed lot. Add a fourth for a reorder programme - the comparison against the retained reference - and the plan starts catching drift as well as defects.

Finally, decide what the inspection is for before commissioning it. A plan written to protect a shipment and a plan written to protect a programme look different, and the second is the one a reorder needs, because the question is not whether this lot passes but whether the product has moved.

Frequently Asked Questions

People Also Ask