When a Sample Is Worth the Cost — and When It Is Not
Samples are usually requested to answer "is this supplier any good". That is the wrong question for a sample to answer, because a factory will always hand-pick the sample.
The questions a sample can answer reliably: does the geometry match the drawing, does the surface finish meet the spec, does the material certificate match the part, and does the packaging survive shipping. These are factual checks against a specification.
The questions a sample cannot answer: whether production units will match, whether they can hold tolerance at volume, whether they will ship on time. Those need either a pre-shipment inspection on the actual batch or a trial order.
So the useful pattern is: request a sample only when there is a specification to check against, and treat a passing sample as permission to place a trial order rather than as approval of the supplier.
For commodity parts where the specification is fully standardised — a 6204 bearing, an M8 grade 8.8 bolt — a sample adds little. The standard defines the part, and certification plus a mill certificate tells you more than a single unit in your hand. Sample fees there are mostly logistics cost.
For custom parts, a sample is close to mandatory, but request it as a first-article inspection with a dimensional report rather than a loose part. A CMM report with measured values tells you whether the process is capable; a part in a bag tells you it exists.
One practical note on cost: expect to pay for samples and shipping, and expect the fee to be credited against the first order if you proceed. A supplier who offers free samples on a custom part is either very confident or pricing the risk into your quotation.