The Real Cost of GB/T 2828.1 AQL Sampling at 3 Levels: 0.65 / 1.5 / 2.5 — When a Printer Quotes 32 Pieces, Which Level Should Be Agreed with the Customer First?

A client in the cultural and creative products industry ordered 32 tea canisters and specified the following acceptance requirements in writing: "Inspect in accordance with GB/T 2828.1, AQL 0.65."

This term is very common in the industry, but it has a major problem: AQL 0.65 is not a complete sampling plan. It is merely a figure in the “Acceptance Criteria” column. To determine the sample size for these 32 units and the number of nonconforming units permitted, two additional inputs are required—the lot size range and the inspection level.

The customer did not specify these two items. We selected them on their behalf according to standard practice: general inspection level II. But this choice directly determines everything that follows.

First, spread out this table

The core logic of GB/T 2828.1 is: first, identify the lot-size range based on the lot size; then, find the sample-size code according to the inspection level; and finally, find the corresponding acceptance/rejection number based on the AQL value.

For a lot size of 32 pieces, general inspection level II falls within the lot size range of 9 to 150. The corresponding sample size code is K, with an AQL of 0.65, an acceptance number Ac=1, and a rejection number Re=2. In other words: Inspect 13 pieces; the lot passes if no more than 1 is defective, and the entire lot is rejected if 2 are defective.

What if the customer requires AQL 2.5? For the same lot size and inspection level, the sample size is still 13 pieces, but the acceptance number becomes Ac=4 and Re=5. The sample size remains unchanged, while the acceptance threshold is four times wider.

This is the part most prone to misunderstanding. Many people think that AQL 0.65 is 4 times stricter than 2.5, whereas it is the "permissible proportion of nonconforming units" that is stricter, not the "amount of work."

What Really Distinguishes the 0.65 / 1.5 / 2.5 Options

The difference lies not in individual inspection, but in three things:

First, the cost of reworking.Out of 32 pieces, 2 were judged defective, resulting in rejection of the entire batch. The printing plant must either inspect all pieces and redo the work, or negotiate with the customer for acceptance under concessions. Inspecting all 32 can bodies would require labor that may be entirely unnecessary when applying AQL 2.5.

Second, the strictness of defect classification.An AQL of 0.65 generally means that the customer has included “minor defects” within the assessment criteria. Minor scratches and extremely minor spots on printed materials are usually handled under a separate Ac/defect rule at the standard inspection level. However, if the customer insists that all defects be treated as critical defects, 0.65 represents a very tight limit.

Third, the burden of proof in the event of a dispute. The higher the AQL, the harder it will be for the customer to claim afterward that “you failed to meet the standard.” An AQL of 0.65, however, gives the customer ample room to identify one minor defect during inspection and reject the entire shipment.

A Real Quotation Process

For the cultural and creative client with 32 tea canisters, we ultimately settled on AQL 1.5.

When I spoke with him, I cited three figures. First, for a lot of 32 pieces, a sample of 13 pieces gives an inspection ratio of 40%, which is already substantial. Second, at AQL 0.65, 2 defective pieces mean rejection of the entire lot, while cultural and creative customers with small batch requirements usually place repeat orders, and any delay in the repeat-order cycle means missing the seasonal sales period. Third, this model’s jar is dark matte, and light-colored scratches are essentially invisible to the naked eye under normal lighting; it is not worth bearing the risk of rejection for defects that cannot be seen.

The customer responded to our inquiry in very practical terms: “We’re not sure it can be detected at 0.65 either. The people inspecting the goods are part-time warehouse staff, not professional QC.”

This statement basically sets the tone. AQL is intended for parties with adequate inspection capabilities. When the inspecting party lacks such capabilities, the lower the AQL, the greater the risk—because it also turns the variable of “whether the defect can be detected” into a risk.

4 Things to Confirm with the Customer Before Quoting a Small-Batch Unit Price

Based on the pitfalls we have encountered, these 4 items must be stated in the quotation; verbal explanations do not count.

First, the inspection level: general inspection level I / II / III or special inspection level.For special inspection levels, the sample size can multiply several times when the lot size is large, resulting in substantially different inspection labor and cost.

Second, defect classification: how are critical, major, and minor defects distinguished, and what are the criteria for each category? If the sentence “Does color deviation count as major or minor?” is left unclear, the printer and the customer will each advance their own interpretation during inspection.

Third, the AQL level, clearly stating whether it is based on “single-lot inspection” or “moving-average inspection across consecutive lots.” Many customers are unaware that moving-average inspection is also an option. This method allows one lot to be slightly looser and the next lot slightly tighter, with the long-term results reflecting the true data.

Fourth, the handling method after rejection: full inspection, rework, acceptance by concession, or re-production, and who bears the respective costs. If this clause is not included, the day of rejection will be the day of the argument.

The One Thing Most Likely to Be Missed in Small-Batch Production

With a lot of 32 pieces, sampling 13 pieces already represents 40%. Inspection takes up a significant amount of labor time, but the greater complication is the lower-limit constraint on sample size.

For batch quantities of 9 to 15 units, the sample size may be only 5 to 8 units. In this case, the significance of the AQL acceptance number is highly limited—allowing 1 nonconforming unit in a sample of 8 units is equivalent to a 12.5% nonconforming rate, which is less stringent than 2.5.

In other words, in the low-volume range, the distinctions between AQL levels are erased by the minimum sample size. When customers insist on AQL 0.65 for low-volume orders, the results often are not significantly different from those of AQL 2.5, yet communication costs double.

For orders like this, we generally advise customers directly: either accept full inspection (we will quote the labor cost for inspecting all 32 pieces accurately), or relax the AQL to 2.5 and agree to “100% sorting for specific defects.” The latter is usually acceptable to both parties.

Practical Use of “Continuous Batch Moving Average”

If the customer is a quarterly customer who places orders consistently, we instead recommend using a moving average.

The logic is simple: applying 100% inspection to a batch of 32 pieces results in a high false-rejection rate. Whether the criteria are strict or loose, either approach is likely to cause disputes. With moving-average inspection, decisions are based on the cumulative results of the most recent batches, so a slightly loose result for one batch is acceptable; only an unfavorable trend triggers rejection.

The practical benefit for the printing plant is that a disputed single-item defect will not hold up delivery of the entire batch. This assumes that both parties can see the cumulative data, which requires the printing plant to properly retain defect records for each batch.

Further Reading

Sampling inspection is only one part of the quality system. Read together with these articles, it provides a more complete picture: