Sample approval: what each stage of sample really proves

Each sample stage answers one question

Early prototypes, often made by a different process entirely, answer whether the form and fit work. Soft-tool or bridge samples answer whether the design is manufacturable and roughly what the process will do. Parts off production tooling answer whether the intended process holds the drawing. Pre-production parts made at rate answer whether it holds under real conditions with real operators. Naming which question each round is meant to settle prevents the recurring failure where a programme treats a favourable early result as clearance to commit tooling and volume. Write the question and the decision criteria into the launch plan alongside the date, so the round cannot be quietly reinterpreted when results arrive.

The appearance sample and the argument it prevents

Colour, gloss, grain, weld appearance, surface blemish and cosmetic acceptance cannot be settled by a written specification alone, because the words mean different things to different people under different lighting. A physical reference agreed by both parties, with matched limit samples showing the worst still acceptable, ends most of these disputes before they start. Store one set at each site, control them like any other reference, and replace them before they fade or discolour. Cosmetic arguments consume a disproportionate share of supplier quality time and this is the cheap remedy.

Sample lead time is programme time nobody schedules

Between requesting a sample and having a decision sit tool manufacture or modification, the run itself, transport, measurement, review and often a second round after findings. That elapsed period is routinely underestimated in launch plans, and the schedule then absorbs it by shortening the approval step rather than the ones before it. Plan sample rounds with realistic intervals including a probable iteration, and be explicit that compressing them means approving on less evidence, so the decision to accept that risk is taken openly rather than by default. Suppliers will usually give an honest estimate of the iteration risk if asked before the first round rather than after it.

Approving with a deviation, and closing it later

Samples often meet almost everything. Where a nonconformance is genuinely acceptable in the application, record a deviation naming the characteristic, the value accepted, the reason it is tolerable and how long it applies. Where it is not acceptable, say so plainly, because a sample approved to keep a programme moving becomes the standard the supplier will produce to forever. Open deviations need a review date and an owner, otherwise they accumulate into a shadow specification nobody has ever seen assembled in one place. Keep a single register of live deviations per part and review it before any design change is released.

What the supplier learns from a sample round

The exchange runs both ways. A supplier producing early samples discovers where the process is difficult, which features fight the tooling, how long the cycle really is and where scrap concentrates. That information is valuable to you and suppliers do not always volunteer it. Ask what they found hard, what they had to adjust, what scrap rate the sample run produced and what they would change in the design. A supplier that reports a flawless first sample run has either done this many times or has not measured much. Their answers also predict where series production will drift, which is worth knowing before you write the control plan.

Frequently asked questions

Why can a prototype pass and the production part fail?

Because they are made differently. Prototypes are often produced by another process altogether, hand-finished, measured selectively and made by skilled staff paying close attention to a single piece. Production parts come off dedicated tooling at rate, with material and cycle time as they really are, made by operators running many parts. Every source of variation that matters in series supply is absent from the prototype, which is why it answers the design question and not the manufacturing one.

How many rounds of samples should a new part need?

Plan for more than one and be pleased if you need fewer. A first round on production tooling almost always produces findings, and the second round demonstrates the corrections. Needing repeated rounds beyond that points at something structural: a drawing the supplier cannot hold, a tolerance derived from a template rather than a function, or a process route that was never suitable. At that point another sample round is not the answer and a technical review of the part is.

Should samples be measured by the supplier, by you, or by an independent laboratory?

The supplier measures and reports, because it needs that discipline for series production. You verify enough of the critical characteristics independently to be confident their measurement agrees with yours. Where the tolerance is tight relative to measurement uncertainty, or where the two sets of results disagree, an independent laboratory settles it. Accepting supplier data without ever checking it means you are trusting a measurement system you have never assessed.