Syntha Lab RU

AQL and batch quality inspection

A sampling-inspection rule: the lot size sets the sample size and the number of allowed defects, and the lot is then accepted or rejected. It protects the brand and the retailer from hidden defects and gives the factory shared terms for a dispute: accept the lot, send it back for rework, or demand a discount.

Production, logistics, channels Brand and retail Verified: 2026-11-14

Open interactively →

Why this document

Inspecting every item is expensive and inspecting none is risky, hence a sample with an agreed acceptance level. Quality control, purchasing and the factory read it: the inspector draws items and counts defects, purchasing decides whether to accept the lot before payment and shipment. If the acceptance level is not in the contract, the factory may call any figure acceptable, and the brand learns about defects from customers. The AQL must therefore be agreed in advance, separately for critical, major and minor defects.

At a glance

What is this document and why is it needed?
A sampling-inspection rule: the lot size sets the sample size and the number of allowed defects, and the lot is then accepted or rejected. It protects the brand and the retailer from hidden defects and gives the factory shared terms for a dispute: accept the lot, send it back for rework, or demand a discount.
Who owns the document and how often is it rebuilt?
Run by the brand’s quality or production manager and the independent inspector; the batch report is signed by the inspector and the factory before shipment; the result goes to buying and finance (payment, penalty, refusal).
What mistakes are the most common?
Treating AQL 2.5% as a guarantee of “no more than 2.5% defects” → a lot with 4–5% defects passes → agree on re-sorting and a second inspection when exceeded. Mixing defect classes into one limit → a critical defect dissolves among minors → separate table and AQL for each class. Drawing the sample from one carton → the result does not represent the lot → random sampling across all cartons and sizes.

For whom

Brand
The brand inspects the batch at the factory before paying the balance and shipping: AQL decides whether the batch can leave and is written into the contract with the defect classification and the re-sorting procedure. Compare the cost of a third-party inspection with the cost of re-sorting and returning from your own data: Lab gives no normative benchmark.
Retail
The retailer applies AQL at warehouse receiving and in contract terms with brands: acceptable defect level, procedure for acts and returns, claim deadlines. For its own imports it does what a brand does; for bought-in goods it runs sampling at receiving plus return rules in the contract.

What it consists of

1. Defect classification

  • Critical: dangerous to people or illegal (sharp parts, needles, prohibited substances, wrong composition); acceptable level 0.
  • Major: reduces usability or appearance, the customer will return it or not buy (hole, stain, wrong size, broken zip).
  • Minor: a deviation the customer probably will not notice (loose thread, uneven stitch in a hidden place).

2. Sampling table (general level II)

  • Lot 2–8 units: code A, sample 2; 9–15: B, 3; 16–25: C, 5; 26–50: D, 8; 51–90: E, 13; 91–150: F, 20; 151–280: G, 32; 281–500: H, 50.
  • 501–1,200: J, 80; 1,201–3,200: K, 125; 3,201–10,000: L, 200; 10,001–35,000: M, 315; 35,001–150,000: N, 500; 150,001–500,000: P, 800; above 500,000: Q, 1,250.
  • Other levels: I (smaller sample, cheaper), III (larger sample, stricter); special levels S-1…S-4 for destructive or costly checks.

3. Acceptance numbers (normal inspection, single plan)

  • Sample 50: AQL 1.5 — accept up to 2, AQL 2.5 — up to 3, AQL 4.0 — up to 5.
  • Sample 80: AQL 1.5 — up to 3, 2.5 — up to 5, 4.0 — up to 7; sample 125: 1.5 — up to 5, 2.5 — up to 7, 4.0 — up to 10.
  • Sample 200: 1.5 — up to 7, 2.5 — up to 10, 4.0 — up to 14; sample 315: 1.5 — up to 10, 2.5 — up to 14, 4.0 — up to 21. Reject when the defect count is one above the acceptance number.

4. Inspection severity

  • Normal inspection by default; tightened after rejections (two of five consecutive lots rejected), reduced after a long run of accepted lots.

5. Inspection report

  • Retailer: Lot, order, quantity, date, level and AQL by defect class, sample size, defects by type, photos, measurements in the sample, verdict, factory and inspector signatures.

What each side fixes

  • Brand: writes into the contract the inspection level, AQL by class, the packed share at inspection and the re-sorting procedure; decides whether to pay the balance before or after inspection.
  • Retailer: on warehouse receipt applies the sample to the brand shipment under the contract terms, counts units not cartons, and issues the act and return within the claim period.

Brand and retailer: what differs in practice

  • Brand: chooses AQL by risk level: critical 0, major 2.5%, minor 4% as an illustrative scheme; fixes the values in the contract and the quality annex.
  • Retailer: receives at the warehouse under the sampling plan and acts: a result above the acceptance number gives a claim to the supplier within the contract term.
  • Retailer: for a small lot and a costly unit uses tightened inspection or a full check: sampling is not justified.
  • Retailer: decides whether to accept a lot with a discount or with sorting: compares the sorting cost with the lost margin.
  • Retailer: keeps a receiving log by supplier and uses the share of rejected lots as a quality indicator when choosing suppliers.
  • Retailer: orders an independent factory inspection before shipment if the supplier is new or the region is distant and compares the inspector report with the receipt.
  • Retailer: links defects on the shelf and customer returns to the supplier and lot so that returns become the basis of the future inspection plan.

How to read the result and what decisions follow

  • Look at the whole curve, not one point: a plan that accepts a lot with a defect rate twice AQL more often than you need is too lenient.
  • If more than the acceptance number is found, the decision is reject, sort or accept at a discount; compare sorting cost with the lot margin.
  • A critical defect (safety) is allowed only at zero: one finding changes the decision regardless of other results.

Control checks

  • The sample size matches the lot size code and inspection level and the acceptance number is taken from the table for the chosen AQL (check against the standard text).
  • The number of defects found in the act equals the sum by category and the decision in the act matches the rule found no more than Ac.

Where the data comes from

Data sourceFields
Contract and specificationlevel and AQL by defect class, classification, re-sorting procedure, penalties
Order and packing listlot size, cartons, sizes and colors
Inspectiondefects by type, photos, measurements, verdict
Finance and buyingunit price, re-sorting cost, balance amount, inspection cost

How to build it step by step

  1. Brand: Step 1: Fix in the contract the inspection level, AQL per class and defect definitions with photo examples.
  2. Brand: Step 2: Schedule the inspection when at least 80% of the batch is packed (state the share in the contract); agree the date in advance.
  3. Brand: Step 3: Determine the sample size from the table based on total units in the lot, not on one carton or style.
  4. Step 4: Draw the sample randomly across cartons and sizes, proportional to sizes and colors.
  5. Step 5: Inspect the items, measure part of the sample against the chart, check labels, packaging and marking; record defect type and class.
  6. Brand: Step 6: Compare the defect count with the acceptance number for each class separately; any class outside the norm rejects the lot.
  7. Brand: Step 7: On rejection the factory re-sorts 100% of the lot and brings a report; then a repeat inspection at the factory’s cost.
  8. Brand: Step 8: Pass the result to finance: payment, penalty, discount or refusal; record it in the factory quality log for the scorecard.

Formulas

MeasureFormulasNote
Probability of accepting the lotPa = Σ for k from 0 to Ac of C(n,k) · p^k · (1−p)^(n−k); n is the sample, p the defect rate in the lot, Ac the acceptance numberIf the sample exceeds 10% of the lot use the hypergeometric law. For bad lots (expected defects n·p above the acceptance number) the binomial formula overstates the acceptance probability and for good lots it understates it: the hypergeometric distribution has the smaller variance.
Expected defective units at the customerDefects at the customer ≈ Pa × p × lot sizeShows that AQL 2.5% does not guarantee at most 2.5% defects: a poor lot is accepted with noticeable probability.
Re-sorting costCost = lot size × cost of checking one unitIf re-sorting is cheaper than a return it is better to demand it at the factory.
Sample shareShare = n ÷ N × 100Shows how much of the lot you saw: for a lot of thousands of units a small share is checked, the rest is accepted on statistics.

Formulas in notation

Probability of acceptance (binomial)
P_a(p) = Σ_{k=0}^{Ac} C(n, k) × p^k × (1 − p)^(n − k)
n sample, Ac acceptance number, p true defect rate. At p = 3% and Ac = 7 we get 96.5%.
Producer risk
α = 1 − P_a(AQL)
The probability of rejecting a lot whose defect rate equals AQL: 1.4% for major defects.
Consumer risk
β = P_a(p_bad)
The probability of accepting a lot with an unacceptable defect rate; falls as the sample grows.
Sample fraction and sorting cost
f = n / N; S = N × c
N lot, c sorting cost per unit. Example: 5.2% and RUB 33,600.

Worked example

Acceptance of a lot of 2,400 units: sample, acceptable defects and verdict

Defect classAQLSampleAccept up toReject fromFoundVerdict
Critical (safety)0%125010accept
Major2.5%125786accept
Minor4.0%125101111reject

A lot of 2,400 units at level II falls under code K, sample 125 (5.2% of the lot). Majors found 6 against a limit of 7: accept. Minors 11 against a limit of 10: one above the allowed number, the class fails and the whole lot is rejected on minors and sent for re-sorting. For a lot with a true major defect rate of 3% the acceptance probability is 96.5%; at exactly the AQL (2.5%) it is 98.6%. Re-sorting 100% of the lot costs RUB 33,600 at RUB 14 per unit: the benchmark against which the inspector day of RUB 18,000 and the return risk are compared.

Figures

Operating characteristic (OC) curve: probability of accepting a lot at the true defect rate
Line chart: Major: sample 125, Ac = 7: 100 → 6.01; Minor: sample 125, Ac = 10: 100 → 28.4020406080100246810AQL 2.5%Major: sample 125, Ac = 7, 0.5: 100Major: sample 125, Ac = 7, 1: 100Major: sample 125, Ac = 7, 1.5: 99.9Major: sample 125, Ac = 7, 2: 99.6Major: sample 125, Ac = 7, 2.5: 98.6Major: sample 125, Ac = 7, 3: 96.5Major: sample 125, Ac = 7, 4: 87.1Major: sample 125, Ac = 7, 5: 71.2Major: sample 125, Ac = 7, 6: 52.2Major: sample 125, Ac = 7, 8: 20.9Major: sample 125, Ac = 7, 10: 6.01Minor: sample 125, Ac = 10, 0.5: 100Minor: sample 125, Ac = 10, 1: 100Minor: sample 125, Ac = 10, 1.5: 100Minor: sample 125, Ac = 10, 2: 100Minor: sample 125, Ac = 10, 2.5: 100Minor: sample 125, Ac = 10, 3: 99.9Minor: sample 125, Ac = 10, 4: 98.8Minor: sample 125, Ac = 10, 5: 95.1Minor: sample 125, Ac = 10, 6: 86.9Minor: sample 125, Ac = 10, 8: 58.3Minor: sample 125, Ac = 10, 10: 28.4%
  • Major: sample 125, Ac = 7
  • Minor: sample 125, Ac = 10
Data as a table
MeasureMajor: sample 125, Ac = 7Minor: sample 125, Ac = 10
0.5100100
1100100
1.599.9100
299.6100
2.598.6100
396.599.9
487.198.8
571.295.1
652.286.9
820.958.3
106.0128.4

Operating characteristic (OC) curve: probability of accepting a lot at the true defect rate. The higher the true defect rate, the lower the chance to accept the lot; a plan with a larger acceptance number is more lenient. The marker shows the AQL of major defects. Binomial approximation: when the sample exceeds 10% of the lot use the hypergeometric calculation. The chosen plans are illustrative, check against the standard text.

Acceptance number Ac by AQL and sample size
Heat map: 3×5; min 2, max 2150508080125125200200315315AQL 1.5%AQL 1.5%2AQL 1.5% / 50: 23AQL 1.5% / 80: 35AQL 1.5% / 125: 57AQL 1.5% / 200: 710AQL 1.5% / 315: 10AQL 2.5%AQL 2.5%3AQL 2.5% / 50: 35AQL 2.5% / 80: 57AQL 2.5% / 125: 710AQL 2.5% / 200: 1014AQL 2.5% / 315: 14AQL 4%AQL 4%5AQL 4% / 50: 57AQL 4% / 80: 710AQL 4% / 125: 1014AQL 4% / 200: 1421AQL 4% / 315: 21
  • 2 … 21 pcs
Data as a table
5080125200315
AQL 1.5%235710
AQL 2.5%3571014
AQL 4%57101421

Acceptance number Ac by AQL and sample size. Rows are AQL, columns are sample size; each cell holds the number of defective units allowed for acceptance.

Lot decision: found and allowed
Bar chart: Critical 0, Major 7, Minor 10024681012CriticalCriticalAllowed (Ac): 00Found: 00MajorMajorAllowed (Ac): 77Found: 66MinorMinorAllowed (Ac): 1010Found: 1111pcs
  • Allowed (Ac)
  • Found
Data as a table
MeasureAllowed (Ac)Found
Critical00
Major76
Minor1011

Lot decision: found and allowed. If more than the acceptance number is found the lot is rejected or sorted; a critical defect is not allowed at all. Lot 2,400 pcs, code K, sample 125 pcs (5.2% of the lot). Sorting the whole lot costs RUB 33,600.

AQL sampling: scheme. A sample is drawn from the lot, defects are counted and compared with the limit: accept or reject.

Specifics by category

Apparel
Apparel: most common major defects are in sewing (stitching, skipped stitches, stains, measurements outside tolerance) and labeling; AQL 2.5 for majors and 4.0 for minors is common practice, but values are fixed in the contract.
Footwear
Footwear: checks include pairing, symmetry, glue and sole attachment, odor, fit on the last; the sample is counted in pairs; sole separation and sharp nails inside are critical. AQL for majors is usually stricter.
Accessories
Accessories: hardware, rivets, fasteners, leather odor, handle strength; for costume jewelry add checks for nickel and sharp edges. No sizes, so sampling is simpler but functional tests matter.
Kids
Kids: critical defects extend (small parts, cords, sharp edges, chemicals); AQL for criticals is zero and stricter than adult apparel for majors.
Men
Menswear: emphasis on symmetry, sleeve and trouser length, fit; classic garments are also judged on interior finishing.
Women
Womenswear: more finishing and decorative details (embroidery, stones, draping), so a higher share of appearance defects and a separate item for color within the lot.

Common mistakes

  • Treating AQL 2.5% as a guarantee of “no more than 2.5% defects” → a lot with 4–5% defects passes → agree on re-sorting and a second inspection when exceeded.
  • Mixing defect classes into one limit → a critical defect dissolves among minors → separate table and AQL for each class.
  • Drawing the sample from one carton → the result does not represent the lot → random sampling across all cartons and sizes.
  • Inspecting before the lot is packed → a different lot ships → inspect at 80–100% readiness and seal the cartons.
  • No procedure in the contract after rejection → a drawn-out dispute → fix re-sorting at the factory’s expense, a repeat inspection and a deadline.
  • Acceptance probability is computed by the binomial when the sample exceeds 10% of the lot → for bad lots acceptance looks likelier than it is → compute with the hypergeometric law.

Russian specifics

  • Contractual penalties for defective goods are governed by the Civil Code (general provisions on contract and penalty, supply contract); amounts and procedure are fixed in the contract. Rule: Civil Code part 1: Civil Code, part one (including Arts. 330, 424).
  • Safety and composition requirements for light-industry goods: TR CU 017/2011; a critical-class violation (composition, hazardous substances) carries liability regardless of AQL. Rule: CU Commission decision 876 of 09.12.2011: TR CU 017/2011 on safety of light-industry products (Customs Union Commission Decision of 09.12.2011 No. 876).
  • The sampling table and acceptance numbers follow ISO 2859-1 (adopted in Russia as GOST R ISO 2859-1); check the values against the standard text: in Lab they are given for reference and not verified against the act text.

Open interactively →