Samples and Incoming Inspection 2026: The Golden Sample, What to Measure, and What a Sample Cannot Tell You

Almost every serious problem in an imported hardware order is a translation failure rather than a manufacturing failure. Two competent parties agreed on words, and the words meant different things.

A sample is the only tool that catches that before it becomes a container. It converts a description into an object, and an object cannot be misunderstood in two ways at once.

THE THREE KINDS OF SAMPLE

A stock sample is an existing product taken off the shelf. It shows you the model, the finish quality and the general build. It is fast and cheap and it is the right thing to ask for when you are evaluating a supplier rather than a specification.

A pre-production sample is made to your specification before the run starts. It is the one that matters when anything differs from the catalogue — a finish, a backset, a spindle length, a printed box — because it proves the specification was understood, not just received.

A production sample is drawn from the actual run. It proves what is in the cartons, and it is the one an inspection is carried out against.

The common mistake is treating a stock sample as though it were a pre-production sample. A stock sample in a stock finish tells you nothing about the custom finish you ordered, and approving one is approving a different product.

THE GOLDEN SAMPLE

This is the procedure that turns a sample from a courtesy into an instrument, and it costs almost nothing.

Two identical samples are prepared. Both are signed and dated by both parties. One is sealed and retained by the buyer; the other is sealed and retained by the factory. Both carry a reference number that appears on the purchase order.

From that point, "as per golden sample 2026-0147" is a specification line, and a dispute about finish or fit is resolved by opening two boxes rather than by exchanging adjectives. It is the single most effective quality control measure available to a buyer who is not standing in the factory.

Retain yours until the order has shipped, been received and been accepted. A golden sample discarded after approval is a golden sample that does not exist when it is needed.

WHAT TO MEASURE WHEN A SAMPLE ARRIVES

Measure against the drawing, not against your memory of the drawing, and record the readings rather than only the pass or fail.

For a lock: backset, centre distance between spindle and cylinder, forend length and width, case dimensions, spindle square, and the door thickness range.

For a handle: lever length from the spindle axis, projection from the door face, rose or plate footprint, spindle square and centre, and the return geometry.

For a hinge: leaf dimensions, thickness, knuckle length, pin diameter, fixing hole positions and countersink type.

For a closer: body dimensions, fixing centres, arm length, and the mountings the arm set supports.

For glass fittings: the glass thickness range, the glass gap, and every cut-out dimension including corner radii.

For all of them: the finish, against a physical reference rather than a photograph or a code. As our finish code page explains, the same code from two suppliers can be visibly different.

THE THREE THINGS A SAMPLE CANNOT SHOW YOU

Being honest about the limits of a sample is what makes the rest of the procedure credible.

Batch consistency. One sample is one sample. It tells you what the factory can make; it does not tell you what the factory will make across five thousand pieces. That is what an inspection with a sampling plan is for.

Corrosion resistance. You cannot see 240 hours of salt spray. A sample that looks perfect and a sample that will streak in a year are indistinguishable in the hand, which is why the corrosion grade has to come from a test rather than from an impression.

Cycle life. A lock that turns sweetly at ten cycles tells you nothing about its condition at two hundred thousand. Durability is a test result, and a sample is not a test.

Anyone who tells you a sample settles these three is either mistaken or selling.

WRITING AN INSPECTION STANDARD SOMEBODY ELSE CAN APPLY

If you use a third-party inspector, the quality of the inspection is set entirely by the quality of the instruction, because the inspector does not know your building and will not phone you from the factory floor.

Name the golden sample reference. State the measurements to be taken and the tolerance on each — a tolerance is what converts a measurement into a decision. State the sampling plan: how many cartons opened, how many pieces measured, and what constitutes a fail.

State what is checked functionally and how many times. "Operate the lock" is not an instruction; "operate the lock ten times and confirm the latch fully retracts and the lever returns to horizontal each time" is.

State the packing checks: carton markings, quantity per carton, inner packing, and whether the carton dimensions match the packing list, because a carton that is not the size the packing list says will not load the way the shipment was planned.

And state what happens on a fail — hold the shipment, re-inspect after rework, or accept with a concession — because an inspector who finds a problem and has no instruction will do nothing, which is the worst of the three outcomes.

WHEN TO TAKE THE SAMPLE

While the specification is still open. This is the timing point that most affects the outcome and it is the one most often got wrong.

A sample taken after the deposit is paid and the run is scheduled is a sample whose only possible outcomes are relief and an argument. A sample taken while the specification can still be edited is a conversation, and a misunderstanding found there costs an email.

As our lead time page sets out, the production clock starts at order confirmation, so sampling early does not delay the order. It moves work into the period that is under your control anyway.

WHAT WE DO

We prepare samples to a written specification and we state on the sample record which specification it was made to, so that an approval is an approval of something identifiable rather than of an object in a box.

Where a sample differs from the catalogue in any respect, we say so in writing with the sample rather than letting the difference be discovered. A supplier who ships a sample that is quietly better than the production intent has not won an approval; they have set a standard they will fail against, which is a worse outcome for both parties than an honest sample and a short conversation.