The Sample Was Right and the Production Was Not

A sample is approved, the order is placed, and six weeks later the cartons arrive with mugs that are recognisably the same product and visibly not the same object. The colour is slightly deeper, the handle sits a degree or two differently, the rim feels thinner.

Products in this guide: Colorful Mini Ceramic Bowls · Porcelain Kids Bowl

Buyers usually read this as carelessness. It is more often arithmetic: the sample was made in a workshop, by hand, one at a time. The production was made on a line, in a mould, in a kiln packed with hundreds of other pieces. Those are different processes, and they produce different objects.

See the full ceramic sourcing process map.

What a sample actually is

Samples in this industry come from three different places, and buyers rarely ask which.

A workshop sample. Made by a technician, often hand-cast or hand-finished, glazed by dipping, fired alongside other samples. This is what most buyers receive when they ask to see a new shape, and it is the version that looks best.

A pilot sample from production tooling. Made on the actual mould and the actual line, in a small run. Closer to the truth, slower to produce, and less commonly offered.

A retained sample from a previous order. Taken off the line during a bulk run. This is the only kind that is unambiguously a production object.

The distinction matters because a workshop sample and a production run differ in almost every variable that determines appearance: forming method, glaze application, kiln position, decoration technique and inspection intensity.

Five sources of the gap

Forming method. Hand casting and jiggering leave different wall thickness distributions. A hand-finished rim can be tuned by the technician until it feels right. A rim formed in a mould is whatever the mould and the trimming tool produce.

Glaze application. Dipping gives a thick, even coat with a characteristic buildup near the foot. Spraying gives thinner, more uniform coverage and a different surface. A sample dipped by hand and a production batch sprayed automatically will not match, even with the same glaze in the same bucket.

Kiln position. A kiln is not isothermal. Pieces near the burner see more heat than pieces at the door, and the difference shows in glaze melt and colour, especially with reactive and saturated glazes. A sample fired alone often lands in the most favourable position; production pieces fill the whole kiln.

Decoration. Hand-applied decals and hand-painted work vary by operator. Machine application is consistent but sits differently on a slightly different body. Either way, a decoration approved on one body is being applied to another.

Inspection intensity. A sample is looked at closely by several people. A production batch is sampled to a statistical plan. Defects that would have been caught on a sample pass through a lot.

What the buyer can actually control

Four actions at the approval stage reduce the gap substantially, and none of them require technical knowledge.

Ask for a production sample, not a workshop sample. When a shape or decoration is new, ask for a sample made on the production tooling. It adds time; it removes most of the surprise.

Write the process into the approval, not just the appearance. The sealed sample should record body, glaze name or code, decoration method, and the firing condition. A sample without the process attached is an object, not a specification.

Approve a range, not a point. Colour, dimensions and weight should be approved with tolerances. "This colour" is unenforceable; "this colour within a stated difference" is.

Require a first-article confirmation. For the first production batch, ask for photographs and measurements of pieces taken from the start, middle and end of the run, before the balance ships.

The master sample guide explains how sealed references keep batches identical.

Telling a shift from a defect

When goods arrive that are not what was approved, the first question is not who is at fault. It is whether the difference is systematic or individual, because the two have different remedies.

Systematic means the whole batch is offset in the same direction: every piece slightly darker, every rim slightly thicker, every handle slightly lower. This points to a process change — a different glaze batch, a different kiln position, a different trimming tool. It will not fix itself and cannot be sorted out.

Individual means most pieces are correct and some are not. That is a defect distribution, and it is addressed by inspection and sorting, not by changing the process.

The diagnostic is simple: put ten pieces from the shipment side by side with the sealed sample and look for a consistent direction of difference. If they are all off the same way, the process moved. If they scatter around the sample, the process is fine and the sorting was not.

The clause that makes this workable

Two sentences in the specification prevent most of these arguments.

The first defines the reference: the sealed sample held by both parties is the standard against which production is judged, and it records the process used to make it.

The second defines the remedy: where a shipment is systematically outside the approved range, the buyer may require rework, replacement or a price adjustment, and the supplier bears the cost of verification.

Without the first, there is nothing to compare against. Without the second, there is a comparison and no consequence.

The spec freeze guide explains how artwork and packing files are locked between orders.

A sample is a promise about a process the buyer has not seen. The gap between promise and delivery is not usually dishonesty — it is the distance between a workshop and a kiln. Closing it is a matter of asking for the right sample and writing down what it was made of.