You cannot open and test every phone in a thousand-unit shipment, but you still need a defensible answer to "is this lot acceptable?" AQL sampling turns that answer into a calculation instead of a feeling.
Full inspection is the instinct, but at volume it costs more than it returns, and it still misses problems because testers get tired. Sampling trades a tiny amount of uncertainty for speed and a documented standard. For a wholesale lot, the question is rarely "is every unit perfect" — it is "is the defect rate below the line I agreed with the supplier."
AQL is the Acceptable Quality Limit: the worst defect rate you would consider acceptable on average. It is a reference for the sampling plan, not a promise that the lot has zero defects, and not a guarantee on any single unit. A 2.5 AQL is a common working level for consumer goods; tighter levels like 1.0 or 0.65 are used for the defects you care about most. Always state the level in the purchase order so the accept/reject rule is agreed before the goods arrive.
The plan starts with the lot size and an inspection level — general level II is the default — which gives you a sample-size letter code. That code, with your chosen AQL, gives the sample size and the accept/reject numbers. A common example: a lot of around 1,000 units at general level II and AQL 2.5 samples 80 units, accepting on 5 or fewer defects and rejecting on 6 or more. Use the published ISO 2859-1 / ANSI Z1.4 tables as your source, and do not round them from memory.
Sampling only helps if you define what a "defect" is. Sort problems into classes so the accept/reject line is meaningful:
| Class | Meaning | Examples | Treatment |
|---|---|---|---|
| Critical | Unsafe or unusable | Swollen battery, blacklisted IMEI, activation lock, won't power on | Typically reject on a single unit |
| Major | Functional or grade-breaking | Dead screen, camera fault, battery below the agreed floor, grade mismatch | Counted against the AQL |
| Minor | Cosmetic, non-functional | Light scratches within grade, faint housing wear | Not counted, or a separate looser level |
The sample must be random and spread across boxes, not taken from the top layer. Record every sampled unit by IMEI with its result and the date, and test each against the same checklist. A sample that only grabs convenient units proves nothing and will not survive a dispute.
Accept, reject, or renegotiate — and tie the outcome to your claim rights. A passing lot still gets per-IMEI records for traceability. A failing lot triggers a written claim with the sample results attached, and typically re-inspection or a credit, depending on your return terms. The sample result is evidence, so keep it on file. For the full receiving process see how to verify a wholesale lot on arrival, and for what to test refer to the bulk inspection checklist and how to read a QC report. If you outsource the work, choosing a third-party inspector means agreeing on their sampling plan up front.
The tables offer more than one plan. Single sampling makes a decision from one sample; double sampling lets you take a smaller first sample and a second one only if the first is inconclusive, which often means less testing overall.
Multiple sampling goes further, testing in stages. In practice most phone wholesalers run single or double sampling — the choice matters when inspection time and labor are the bottleneck rather than the statistical precision.
Whatever plan you pick, write it into the purchase order so the supplier is judging the same lot by the same rule.
AppleUsed supplies inspected, graded pre-owned iPhones with documented QC and IMEI lists — so your incoming sampling has a supplier that holds up to it.
Request Our Price ListAppleUsed · Pre-owned iPhone wholesale supplier · All guides · Contact