Skip to content
Zilzul — Field notes for fast-moving teams

Sampling levels used in glassware ceramic inspection: part 2

admin About the author
Published
The short answer

Basic sampling plans assume one uniform lot and one verdict. Real orders are messier: mixed designs in one container, re-inspections after failure, and destination markets that each want their own evidence. These situations need sampling decisions beyond the standard table lookup. This second part covers mixed lots, tightened and reduced examination, and aligning samples with market requirements.

Key takeawaysMixed lots need stratified sampling so each design subgroup contributes its own evidence. · Tightened examination answers repeat failure; reduced examination rewards proven stability. · Re-inspections should target the failed checkpoints with a defined sample, not repeat everything blindly. · Marking rules such as the CE framework show why evidence must match the claiming market. [1] · Transparent technical regulations let buyers plan one sampling round that serves several markets. [2]

The first part of this guide explained how lot size and inspection level produce a sample size and how acceptance numbers turn defect counts into verdicts. That machinery handles the standard case well: one design, one shipment, one decision.

Complexity arrives with variety. A single container may hold six mug designs in three colours with two gift-box variants. A failed lot returns for re-examination with only some checkpoints disputed. A buyer splitting shipments between markets needs evidence each authority accepts.

This article addresses each complication with practical sampling responses. The mathematics stays in the background; the focus is on decisions buyers and inspection coordinators actually make.

Sampling mixed designs without hiding failures

Averaging across mixed designs is the most common sampling error in tableware. One failing design hides inside five passing ones, the blended defect rate stays under the limit, and the shipment sails with a systematic fault concentrated in the very cartons one customer will open.

The remedy is stratification: treat each design or colour as a sub-lot with its own sample slice and, where the contract requires, its own verdict. Allocate sample pieces roughly in proportion to subgroup sizes while guaranteeing a minimum slice for small subgroups. Record results per subgroup so the report shows which designs passed rather than merely that the container averaged out.

See the inspection scope for how service stages map to these situations, and state subgroup verdict needs in the booking so the visit is staffed and scheduled accordingly. Inspectors cannot stratify well when they first learn about six hidden variants on arrival.

Tightened, normal, and reduced examination

RegimeWhen it appliesWhat changesTableware example
Normal examinationDefault for new or stable supplyStandard acceptance numbersFirst three orders from a kiln
Tightened examinationRecent failures at this supplierStricter acceptance numbersTwo failed glaze batches in a row
Reduced examinationLong stable historySmaller samplesEighteen passing months of plain tumblers
Return to normalPasses under tightened rulesStandard numbers resumeTwo consecutive passes on probation
Discontinuation riskFailures persist under tighteningStop and fix the processRepeated crazing across re-firings

Switching rules between these regimes should be written into the supply agreement rather than improvised per shipment. Suppliers accept tightened examination as fair when the trigger conditions were agreed in advance; imposed without warning, the same regime reads as punishment and cooperation suffers.

Illustration: Tightened, normal Decorative illustration for the section "Tightened, normal"; visual only, carries no data.

Re-inspection sampling after failure

Re-inspection is not a replay of the first visit. Its purpose is narrower: verify that the agreed remedies worked on the failed checkpoints. A sensible re-inspection draws its sample from the reworked or sorted goods and concentrates examination time on the previously failed attributes while spot-checking the rest.

Define the re-inspection population carefully. If the factory re-sorted the lot and claims a conforming sub-lot, the sample comes from that sub-lot alone and the verdict covers it alone. Mixing reworked and untouched cartons into one sample produces a verdict about nothing in particular.

Document the linkage between visits: what failed, what was agreed, what was done, and what the new sample shows. This chain turns two reports into one intelligible story that buyers, suppliers, and marketplaces can all follow without re-reading the full files.

One sample round serving several markets

Buyers splitting shipments across markets often assume each destination needs its own inspection. Frequently one well-planned visit can serve all of them, provided the checkpoints cover the strictest market's needs and the report records evidence at that granularity.

Illustration: One sample round serving several Decorative illustration for the section "One sample round serving several"; visual only, carries no data.

The logic mirrors how product markings work. The CE marking, for instance, attests conformity with the EU legislation applicable to that product group rather than acting as a universal quality seal. [1] Evidence collected for one market's file does not automatically satisfy another's, so the sampling plan must name each market's checkpoints explicitly instead of assuming overlap.

International disciplines on technical regulation help here by pushing members toward published, transparent requirements. [2] Map each market's documented requirements to checkpoints before the visit, collect the union of evidence in one round, and file market-specific summaries from the shared report. The result costs one visit and satisfies several files.

Booking note for multi-market lots

List every destination market and its special checkpoints in the booking, flag which subgroups serve which market, and ask for per-market summary lines in the report. An hour of booking clarity replaces weeks of follow-up questions.

Sources

  1. Wikipedia: CE marking —— Overview of CE marking as conformity marking for products sold in the European Economic Area.
  2. WTO: Agreement on Technical Barriers to Trade —— WTO portal for the TBT Agreement covering technical regulations, standards and conformity assessment.

Frequently asked questions

Should small subgroups get their own verdict?

When the subgroup ships to a distinct customer or market, yes. A separate verdict prevents a failing variant from hiding in the average. For tiny decorative variants, a minimum sample slice with recorded results is the proportionate answer.

How do I move a supplier to tightened examination?

Invoke the switching rule in your agreement after the agreed number of failures, and tell the supplier which regime now applies. Return to normal examination after the agreed run of passes. Written triggers keep the move factual rather than personal.

Does re-inspection need the full sample size?

Not necessarily. Size it to verify the remedies on the failed checkpoints with statistical sense, while spot-checking the rest. Agree the re-inspection scope in writing before the visit so both sides know what the new verdict will cover.

Can one inspection serve both the EU and other markets?

Often yes, when the visit collects the union of required checkpoints and the report records them separately. Confirm each market's requirements in advance rather than assuming one market's evidence transfers automatically.

Ship a real sprint this week

Stop running standups. Start shipping story points.

Spin up a workspace in 11 minutes — async standups, auto-generated release notes, and cycle-time analytics that actually match DORA. No credit card. Work email only.

Start a free sprint