On this page
Sampling looks like a sequence of increasingly finished versions of the same garment. It is actually a sequence of different questions, and each sample exists to answer exactly one of them. Approving a sample against the wrong question is the most common way a fit error survives all the way into production and then into your returns data.
The Stages and What Each One Answers
Proto Sample
The first physical interpretation of the design. It answers whether the idea works as a garment at all: proportion, silhouette, construction feasibility. It is usually made in substitute fabric and is not a fit sample. Judging fit on a proto is the first classic error, because the fabric behaviour is wrong and the pattern has not been corrected yet.
Fit Sample
Made in correct or near-correct fabric and fitted on a fit model or a form. This is where fit is actually decided. There are usually several rounds, each responding to written comments, and the number of rounds is the strongest predictor of whether the final garment fits as intended. The role of the fit model itself is covered in what is a fit model and why it matters.
Salesman or Photo Sample
Made for showing rather than for approving. It represents the design to buyers or to a photographer and is frequently produced before fit is finalised. Merchandising teams sometimes measure it and publish the numbers, which is the second classic error, because this sample was never intended to be dimensionally correct.
Pre-Production Sample
Made on the production line, in production fabric, with production trims, before the bulk run starts. This is the sample that proves the factory can make what was specified. Approving it is approving the production, and any fit change requested at this point is expensive and late.
Top of Production Sample
Pulled from the first units of the actual bulk run. It proves that what was approved is what is being made. It is the only sample that reflects the garment a customer will receive, and it is the one most often skipped.
| Stage | Question it answers | Correct fabric | Fit decision |
|---|---|---|---|
| Proto | Does the design work? | No | No |
| Fit | Does it fit as intended? | Usually yes | Yes |
| Salesman | How does it present? | Sometimes | No |
| Pre-production | Can the line make it? | Yes | Confirm only |
| Top of production | Is bulk matching approval? | Yes | Verify only |
Where Fit Errors Slip Through
Three patterns account for most of them. Fit judged on a proto, where the fabric is wrong. Measurements published from a salesman sample, which was never dimensionally controlled. And a top-of-production sample skipped under time pressure, which means nobody ever compared the bulk against the approval.
The third is the most consequential because it is invisible. Every earlier stage can be correct and the bulk still drift, particularly when a factory substitutes a component or a second line runs part of the order. Testing standards from bodies such as ASTM exist precisely because material and construction substitutions change behaviour in ways a visual check does not catch.
Making the Sequence Actually Work
Write down which question each sample answers and refuse to answer a different one from it. That single discipline prevents most of the failures above and costs nothing.
Measure every fit sample against the specification rather than against the previous sample, because comparing samples to each other lets error accumulate in one direction. Record the measurements alongside the fit comments so that the next round has both the intention and the result.
Never publish a size chart from a sample. Publish from the specification, and use the top-of-production sample to verify that the specification is being met, an argument developed in how to reduce fit errors before production.
When to Add a Round Rather Than Approve
If a fit sample requires more than a few corrections, the honest answer is another round rather than an approval with comments. Approving with comments transfers the correction to the pre-production stage, where it is more expensive and less likely to be verified.
The temptation to approve is always schedule pressure, and the cost lands later as returns rather than sooner as a delay. Merchants who track fit-related return rate by style usually find that the styles with the fewest fit rounds are the ones generating the complaints, which makes the trade-off visible rather than theoretical.
Frequently Asked Questions
Can I judge fit from a proto sample?
No. A proto is made in substitute fabric to test whether the design works as a garment, and fabric behaviour is a major determinant of fit. Judging fit at this stage produces corrections based on how the wrong material drapes, which then have to be undone once the correct fabric arrives, wasting a round rather than saving one.
Why should measurements never come from a salesman sample?
Because that sample exists to present the design, not to be dimensionally correct, and it is frequently made before fit is finalised. Publishing a size chart from it describes a garment that will not be produced. Size charts should come from the specification, verified against a top-of-production sample.
What happens if I skip the top-of-production sample?
You lose the only check that the bulk matches what you approved. Every earlier stage can be correct while the bulk still drifts, particularly if a factory substitutes a component or runs part of the order on a second line. The failure is invisible until it reaches customers, which is why this is the most consequential sample to skip.
How many fit rounds should a style get?
As many as it needs, which is more than schedule pressure usually allows. If a fit sample requires more than a few corrections, another round is the honest answer; approving with comments pushes the correction to a stage where it is more expensive and less likely to be verified. Tracking fit-related returns by style usually shows that the least-sampled styles generate the complaints.