A supplier writes “AQL inspection passed” on a packing report. That sentence does not identify the lot, the inspected characteristics, the sampling system, the actual sample, or the release decision. It may refer to a cosmetic inspection while the buyer assumes it covers conductor identity, terminal seating and electrical performance. The problem is not solved by adding another percentage to the purchase order. The buyer needs a reproducible decision rule tied to the delivered product.
Direct answer: define the product and lot first, then the inspection unit, nonconformity classes, applicable sampling standard and edition, inspection state, sample selection, test methods and acceptance rules. Require the actual results and a traceable disposition for the same lot. AQL means acceptance quality limit: an index within the specified sampling system. It does not guarantee a maximum defect percentage in every shipment, and an acceptance number of zero does not prove that the uninspected remainder contains no defects.
This guide concerns purchasing and receiving decisions for cable assemblies, terminals, fuse accessories and similar electrical components. It does not select a safety-critical sampling plan or establish a supplier's inspection capability. The responsible quality and engineering teams must approve the applicable product requirements and risk controls. For assembly workmanship criteria, start with the cable harness workmanship acceptance guide. Sampling decides how a defined requirement is evaluated across a lot; it does not supply the missing workmanship requirement.
Separate product approval, process control and lot acceptance
These activities answer different questions. Product qualification evaluates a defined design against applicable requirements. Process control monitors how production behaves. Lot acceptance supports a decision about a submitted group of items using the agreed inspection evidence. None automatically replaces the other two.
A qualification report for a connector family does not show that a particular shipment contains the ordered cavity seals. A process chart does not identify which cartons were selected at receiving. A passed cosmetic sample does not establish a fuse's interruption capability. Each document should state the decision it supports, so the buyer can recognize both relevant evidence and remaining gaps.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Decision | Identity needed | Evidence that does not substitute for it |
|---|---|---|
| Design approval | Exact model, drawing and applicable requirement | An undifferentiated shipment inspection stamp |
| Process monitoring | Process, characteristic, method and time sequence | A single final sample result |
| Lot release | Defined lot, sample and acceptance plan | A certificate for a different product variant |
| Individual unit disposition | Unit identity and observed nonconformity | General acceptance of the surrounding lot |
| Reworked lot release | Original lot, controlled rework and new disposition | A replacement report with the failure erased |
Put the evidence responsibilities into the RFQ. If the seller supplies only loose components, identify which assembly-level checks remain with the integrator. If a finished harness is ordered, define the assembly drawing and required test records. A clear supply boundary prevents one party from assuming that another party has evaluated a critical interface.
Name the sampling system and its edition
Do not write only “standard AQL.” Record the standard identifier, edition and agreed procedure. The official ISO 2859-1:2026 publication page, checked on 10 October 2026, identifies the third edition published in January 2026 and shows that it replaces the earlier 1999 edition and its amendments. That edition identity matters when a legacy supplier form cites an older procedure. The public overview establishes scope and publication status; the actual plan must be selected from the controlled applicable document.
The official ASQ overview of Z1.4 and Z1.9 describes Z1.4 as an attributes acceptance-sampling system for a continuing stream of lots, with normal, tightened and reduced inspection and switching rules. It distinguishes Z1.9, which uses measurement data under stated distribution assumptions. These are not interchangeable labels to attach to the same spreadsheet after inspection.
A contract may intentionally specify an older edition. The purchasing team should identify that requirement and resolve any proposed change through the document-control process. Do not silently change a supplier's sampling system because a newer publication exists. Conversely, do not describe an old edition as the current publication without checking its status.
The RFQ should ask whether the supplier is applying a complete continuing-lot system, an isolated-lot plan, or a customer-specific procedure. A table copied without the associated rules can omit important parts of the decision system. Keep the applicable procedure available to the people selecting samples and authorizing release. Require a completed plan-selection record. Where the chosen system uses them, record lot size, inspection level, AQL for each characteristic class, inspection state, code letter, sample size, and acceptance and rejection numbers. Include the procedure or table reference and any applicable instructions that change the selection. This record makes the supplier's proposal reviewable; it is not a substitute table or a recommendation of values. For a multi-stage plan, record the stage and cumulative decision boundaries before inspection starts.
Define a lot that can actually be traced
A purchase order, a shipment and an inspection lot need not be the same thing. A shipment may contain several part numbers, drawing revisions or production batches. If those distinctions are lost, a sample of the easiest cartons to reach may say little about the material the buyer is about to install.
Define the proposed lot boundary before inspection. Record the part number, revision, quantity, manufacturing or supplier batch identifiers, and any agreed limitations on combining material. Where the responsible quality team permits pooling, retain the evidence for that decision and preserve the underlying identities. A shared delivery date alone is not an adequate explanation for treating different configurations as one homogeneous population.
For a hypothetical receiving example, a delivery contains 1,200 cable assemblies: 700 units of revision C and 500 of revision D. The buyer should not automatically select one sample from all 1,200 and report it against revision D. The quality plan must specify whether separate lots are required, which characteristics differ, and how the selected sample represents the defined population. These quantities illustrate the identity issue, not a standard lot-size threshold.
Keep the physical and electronic lot boundaries aligned. Carton labels, pallet maps, receiving records and inspection reports should let a reviewer reconstruct which units were eligible for selection. If cartons are split, repacked or moved, preserve their relationship to the original lot. A good report should remain interpretable after the inspector is no longer available to explain it.
Decide what is counted before choosing the sample
A nonconforming unit and a count of nonconformities are different measures. A harness might have a wrong label, an omitted clip and a dimensional deviation. Depending on the agreed plan, that could be one nonconforming unit with several recorded findings, or several counted nonconformities. The reporting method must be explicit before anyone compares the count with an acceptance number.
Use a characteristic register with a clear specification, inspection method, classification and counting convention. For dimensions, include the datum and limit, not just “check size.” For markings, specify the required content and legibility criteria. For connector assembly, identify the relevant seating and locking evidence. A category called “appearance” should not conceal multiple undefined technical decisions.
The buyer and supplier should agree how combinations of findings are handled. Define whether a unit with findings in several classes is counted once in each applicable class, how repeated findings are recorded, and which decision takes precedence. Use the controlling procedure rather than inventing a convenient convention after seeing the results.
Product risk determines how critical, major or minor classifications are assigned. These terms should not be copied from a generic consumer-goods checklist into an electrical assembly order without engineering review. A small visible discrepancy can matter if it changes identification or mating, while another cosmetic variation may be outside the functional acceptance requirement. Record the reason for the classification and the authority that approved it.
Make random selection operational
Random selection must describe how an inspector reaches the intended population. “Pick a few from the top” is not a reproducible method when a shipment contains multiple cartons or production subgroups. A practical plan identifies the sampling frame, how items are numbered or located, who chooses them, and how inaccessible selections are handled.
The NIST discussion of lot acceptance plans describes single sampling as a random sample followed by a lot decision, and distinguishes single, double, multiple and sequential approaches. Its overview also separates producer's and consumer's risks. A purchase requirement should preserve those distinctions rather than assume that every plan is “inspect once and pass if nothing obvious appears.”
For packed components, prepare a carton and position map or another approved means of selecting from the defined lot. If a supplier preselects samples before the buyer's inspector arrives, the report should disclose that selection method. Those pieces may be useful engineering samples, but they are not automatically equivalent to a random receiving sample.
If selection is stratified across subgroups, document the design and its intended interpretation. Taking the same number from every carton can give unequal selection probabilities when carton quantities differ. The quality specialist should determine whether that matters for the agreed plan and how to handle it. The buyer does not need to invent the statistics, but should require an accountable selection method.
Read the acceptance number as a decision rule
For a single plan counting nonconforming units, sample size n and acceptance number c summarize the basic decision rule. The rule concerns the observed count under that plan. It does not measure every item in the lot or convert the observed sample fraction into a guaranteed shipment defect rate.
The NIST explanation of operating-characteristic curves shows how acceptance probability changes with the underlying nonconforming fraction. Its binomial calculation assumes that the lot is large relative to the sample. For procurement, the useful lesson is that plans trade different probabilities of accepting or rejecting material; a label alone does not express all of those risks.
The following numbers are an independently calculated, explicitly hypothetical illustration. Assume random selection, independent unit outcomes, error-free classification, an effectively large population, a sample of 50 units and a true nonconforming fraction p. For c = 0, acceptance probability is (1 - p)^50. For c = 1, add 50 × p × (1 - p)^49. These are probability examples, not a prescribed ISO or ASQ table entry, a recommended inspection plan or evidence about any supplier.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| Hypothetical true nonconforming fraction | Acceptance probability with n = 50, c = 0 | Acceptance probability with n = 50, c = 1 |
|---|---|---|
| 1% | About 60.5% | About 91.1% |
| 5% | About 7.7% | About 27.9% |
A zero acceptance number therefore does not make the probability of accepting a non-perfect lot zero. It means that the plan accepts only when the inspected sample contains no counted nonconforming units. The illustration also shows why two proposed plans should not be called equally strict merely because both inspect 50 items. Separate lot acceptance from the disposition of observed nonconforming pieces. In the hypothetical c = 1 case, a count of one can satisfy the lot rule, but the identified nonconforming unit still needs controlled disposition. The RFQ should require its segregation and authorized correction, replacement or other documented decision. Lot acceptance must not silently relabel that known finding as conforming material.
The table does not establish a confidence statement about a particular observed shipment. It starts with assumed population fractions and calculates acceptance probabilities. Finite-lot sampling, clustered defects, selection bias or imperfect measurement can require a different model. Ask the responsible quality specialist to choose and document the plan for the actual risk rather than copying these numbers into a purchase order.
Match the inspection method to the characteristic
A larger sample cannot repair an inspection method that misses the relevant defect. Specify the instrument or fixture, drawing datum, resolution or uncertainty requirements where applicable, environmental conditions, operator instructions and result format. For subjective visual decisions, provide controlled examples and an escalation process for borderline findings.
For harness dimensions, a repeatable fixture and datum may matter more than an extra line on the report. For terminal seating, define the assembly condition and permitted inspection method. For electrical results, identify the test configuration and required acceptance limits. The end-of-line fixture and false-failure guide explains why fixture condition and contact interfaces need their own evidence.
Ask how the supplier distinguishes a product failure from an invalid measurement. The original reading, reason for invalidation and authorized repeat should remain visible. Repeated measurements must not become a way to replace an inconvenient failure with the first passing result. If the method is not valid, resolve the measurement problem and decide what earlier results need review.
Where a measurement lies near a limit, the agreed decision rule should address the applicable uncertainty and specification requirements. Do not create an undocumented rounding convention during receiving. Record units, resolution and the decision basis so two reviewers can understand why the same observation was accepted or rejected.
Send your component lot, inspection characteristics and acceptance-evidence requirements.
Separate destructive checks from routine lot screening
A destructive pull, sectioning operation or material investigation consumes or alters a sample. Agree in advance which pieces will be removed, who bears that cost, whether replacement pieces belong to the original lot, and how their identities connect to the inspection record. A sample selected after convenient pieces have already been removed may no longer follow the intended selection procedure.
A type test or qualification assessment answers a different question from routine receiving inspection. Passing a destructive test on several pieces does not automatically qualify a different material, tool, conductor combination or assembly process. Conversely, an approved design does not remove the need to control manufacturing changes and nonconforming shipments. Specify both sets of evidence when both questions matter.
For a safety-relevant characteristic, do not choose an AQL merely because it appears on a previous commercial order. The governing product requirements, risk assessment and responsible engineering authority must determine the appropriate controls. Some characteristics may require individual verification, validated process controls, qualification evidence or several controls together. This guide does not assign a universal sampling rule to safety-critical defects.
Even a stated “100% tested” shipment needs an understandable method. Ask which characteristic was checked, the test limits, equipment status, item-to-result traceability and handling of failed or untested units. Testing every piece with an unsuitable method is not proof that the shipment meets every requirement. Inspection coverage and measurement capability are separate questions.
Make inspection history part of the purchasing agreement
Where the agreed system uses normal, tightened and reduced inspection, identify the rules and the records needed to apply them. The ASQ overview of Z1.4 and Z1.9 describes switching rules as part of the continuing-lot systems. Do not extract a sample size while omitting the history that governs its use.
The purchase package should identify who maintains the lot sequence, who authorizes a change in inspection status and how the buyer can review that decision. Keep the product and process scope of that history explicit. Good results on one connector family are not automatically evidence for a new terminal material, another production line or a changed assembly revision.
If shipment history is unavailable, say so. Ask the quality owner to select the applicable starting conditions under the agreed system. A new supplier should not simply declare reduced inspection because its previous customer inspected fewer pieces. Equally, do not invent a universal number of passing lots that applies to every standard and edition; use the actual rules selected in the contract.
For an isolated purchase rather than a continuing series, the responsible specialist should confirm that the selected approach fits the transaction and the intended protection. Record the rationale and the relevant risks. The cheapest inspection quotation may assume an ongoing process history that the buyer does not possess.
Define what happens after a rejected lot
Agree the disposition before an inspector finds a failure. A useful workflow separates containment, investigation, authorized rework or sorting, verification and the final release decision. Identify the physical storage status and electronic lot status so rejected material cannot be mixed with accepted stock while the parties discuss the result.
A rejected sample does not necessarily identify the extent or cause of a problem. It triggers the agreed lot decision. Any follow-up screening needs its own method and scope, particularly if the original defect is difficult to detect or may have been introduced across multiple lots. Investigating adjacent shipments may be appropriate, but that decision should follow evidence and the responsible quality process.
Preserve the original inspection result when material is reworked or resubmitted. Record the original lot identity, defect findings, sorting or repair instructions, affected quantity, remaining quantity, revised identity if applicable and the authorization for reinspection. Relabeling rejected material as a fresh lot without that connection removes information the buyer needs to interpret the second result.
Avoid an informal practice of drawing another sample repeatedly until one passes. A double or sequential plan is a predefined decision procedure, not permission to ignore a failed single-sampling result. If the agreed plan allows another stage, follow its actual rules and cumulative accounting. Otherwise obtain an authorized disposition instead of silently changing the plan after seeing the data.
Compare quotations by the evidence they will deliver
Give bidders the same lot definition, characteristic list and proposed acceptance system. Ask them to identify deviations rather than bury assumptions in a generic quality statement. Inspection time, destructive samples, fixtures and reporting can affect the quotation, but no fixed cost or lead time follows from an AQL label alone.
On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.
| RFQ field | Evidence to request | Question to resolve before order release |
|---|---|---|
| Lot identity and quantity | Part and revision list, batch boundaries, packing map | Are different revisions or processes being pooled? |
| Characteristic definition | Drawing reference, defect examples, units and limits | Will both parties classify the same observation consistently? |
| Sampling basis | Named system and edition, plan inputs and decision rules | Is this a continuing series or an isolated lot? |
| Selection and inspection | Selection record, method, equipment and operator basis | Can the procedure detect the specified nonconformity? |
| Results and disposition | Raw counts, unit identities, failures and authorization | Can accepted, rejected and resubmitted material be distinguished? |
A completed report should make the decision reproducible by a second reviewer. It need not expose unrelated proprietary production details, but it should identify the agreed requirements and show how the observations support the disposition. A cover page reading “AQL passed” without the underlying characteristic counts is difficult to evaluate.
Set a practical document handover point: for example, before shipment release or before a defined receiving decision. This is a proposed contract choice, not a promise of any supplier's existing service. Agree who signs, how exceptions are approved and how long relevant records must remain accessible under the purchase terms.
Buyer FAQ
Is AQL the maximum defect percentage allowed in every shipment?
No. AQL is an indexing value used within a specified acceptance-sampling system. It is not a guarantee that every accepted lot has a nonconforming percentage below that number. The plan's operating characteristics, selection method and applicable rules determine its decision behavior. State the complete sampling basis in the purchase agreement.
Does an acceptance number of zero mean the entire lot has zero defects?
No. It means the inspected sample must contain no counted nonconforming units under that rule. Uninspected units remain unobserved. The hypothetical probability example above shows that a lot with nonconforming units can still produce a clean sample. Do not describe sample acceptance as proof of a defect-free shipment.
Can we use the same AQL for every electrical component?
Do not assume that. Product requirements, characteristic severity, process evidence and the selected sampling system affect the decision. Appearance, dimensions and safety-related functions may need different controls. Ask the responsible quality and engineering owners to define the characteristic-specific requirements rather than copying one commercial label across the entire order.
What if the supplier and buyer cite different standard editions?
Resolve the difference before selecting the plan or accepting the report. Record the exact system and edition, and review how the proposed inspection was determined. The ISO public catalogue identifies ISO 2859-1:2026 as a new edition; an older contract is not automatically rewritten merely because a newer edition exists.
Can we combine several part numbers into one inspection lot?
Only when the approved lot definition and sampling approach support that grouping. Different revisions, materials or process conditions can make pooled results misleading. Explain the proposed grouping, record quantities and require the quality owner to approve the basis. Convenience in packing or invoicing is not by itself a statistical justification.
Should an inspector take equal quantities from each carton?
That can be appropriate for a deliberately designed selection procedure, but it is not automatically equivalent to simple random selection of units. Unequal carton quantities may change selection probabilities. Document the sampling frame and method so the result can be interpreted using the intended plan rather than an unstated convenience sample.
Can supplier-selected samples replace receiving inspection?
They can support engineering review, but do not automatically represent a randomly selected receiving sample. Agree who selects the pieces, when selection occurs and which lot they represent. If the buyer accepts another arrangement, document its limits and the evidence supporting it rather than labeling all submitted samples as equivalent.
Is a dimensional sampling plan also suitable for continuity testing?
Not necessarily. The characteristic, measurement or classification method, failure consequence and applicable requirements differ. Specify the continuity test configuration and result criteria separately. A common report template can contain both checks, but the acceptance basis for each must be explicit and technically appropriate for the application.
How should damaged packaging affect the inspection decision?
Define packaging requirements and possible product consequences separately. A damaged carton may require containment or additional investigation even if sampled electrical results pass. Record what was observed and follow the agreed disposition process. Passing one characteristic does not waive another requirement or prove that handling-related damage is absent.
Can a failed lot be accepted after a second sample passes?
Only through the agreed plan or an explicitly authorized disposition. Preserve the original failure and any rework or screening evidence. A predefined multi-stage plan has its own decision rules; repeatedly taking replacement samples until a passing result appears does not provide the same basis for acceptance.
Will a stricter plan automatically increase price or delay delivery?
The effect depends on the actual inspection, sample consumption, fixtures, reporting and production arrangement. Request a specific quotation and identify which requirements drive any difference. This guide makes no price, MOQ, capacity or lead-time promise. Compare evidence and deviations alongside the commercial terms.
What information should we send with an RFQ?
Send the part and drawing revisions, lot quantities and boundaries, critical characteristics, proposed sampling system and edition, inspection methods, evidence format and disposition requirements. Identify unresolved decisions rather than choosing unsupported values. The supplier's response should distinguish confirmed requirements, exceptions and items needing engineering or quality approval.
Prepare a lot-release requirement that can be checked
A useful electrical component acceptance sampling RFQ connects the commercial shipment to a defined population, observable characteristics and an agreed decision procedure. It also explains what inspection does not establish. Keep product qualification, process evidence and lot release connected without presenting them as interchangeable approvals.
Before issuing the order, have the responsible parties review the lot definition, sampling basis, random selection, inspection validity and rejected-lot handling together. Those details make the result interpretable and prevent a short quality label from carrying promises that neither the standard nor the observations support.
Send your drawings, lot definitions and inspection-report requirements for an RFQ review