Why Your Size Medium Fits Differently Lot to Lot (And What's Actually Causing It at the Factory)

If you're getting returns and the customer keeps writing 'sizing is off' or 'fits smaller than expected,' the instinct is to blame the size chart. But nine times out of ten, the size chart is fine. The problem is that the garment didn't match the spec by the time it shipped. Understanding how to maintain consistent sizing across production runs isn't really about charts or guides — it's about what happens on the cutting table and in the sewing room, and whether anyone is actually checking. Most aren't.

Apparel return rates on ecommerce platforms run between 20% and 40%. Across multiple U.S. retailer surveys from 2023 to 2025, fit is cited as the primary reason in roughly 70% of those returns. That's not a marketing problem. That's a production problem. And it's one that starts well before the garment gets packed.

Why Sizing Inconsistency Is a Production Problem, Not a Design Problem

Designers draw a silhouette. They spec a medium, get a sample, try it on, approve it, and move on. What they're often not thinking about is what happens when that sample goes into a full production run across four sizes, on a different roll of fabric, cut by a different marker, sewn by operators working under time pressure. The design was fine. The production is where it falls apart.

This distinction matters because brands spend a lot of energy refining their fit on the sample and very little energy specifying what the factory should do to protect that fit through the full run. The sample is a prototype. The production run is a manufacturing process. Those are two different things, and conflating them is where the trouble starts.

A factory operating at speed, with a 500-unit run across four sizes, is making dozens of micro-decisions about how to interpret your spec. If you haven't given them clear, unambiguous instructions with tolerances, they will make those calls themselves. Some of them will be wrong. And by the time you find out, you're looking at re-cut costs, timeline delays, or worse, you're finding out because your customers are telling you.

What Grading Actually Means and Where Factories Cut Corners on It

Grading is the process of systematically scaling a base size pattern up and down to produce the full size range. It sounds mechanical, like simple arithmetic. And it mostly is, except that a small error compounded across multiple sizes adds up fast.

Here's the specific version of this problem: a grading increment error of just 1.5 cm per size, compounded across a four-size run, puts the top size 6 cm off-spec from the original fit intention. Six centimetres on a chest measurement is visible on the body. That's not a subtle difference. That's a garment that looks wrong, that fits wrong, and that a customer is going to return.

Where do factories cut corners on grading? Usually in one of two places. First, they're grading from a brand-supplied tech pack that wasn't built with proper grade rules. The brand specifies measurements for each size individually, and those numbers weren't generated by applying consistent grade increments — they were filled in somewhat intuitively. Second, the factory's pattern team is working fast, and they apply a generic grade rule that doesn't match your specific silhouette. A fitted bodice and a relaxed tee require different grade rules. Using the same one for both causes fit distortion in different parts of the garment at different sizes.

The ASTM D5585 standard provides body measurement tables for women's misses apparel that U.S. brands can use as a grading baseline. Fewer than 30% of independent brand founders are familiar with it when they submit their first tech pack. If you're not familiar with it either, that's worth fixing before you write another size spec.

The Role of the Fit Sample and Why One Approved Sample Isn't Enough

Most brands submit one approved fit sample and assume it governs the full run. It doesn't. It governs one size, on one fabric swatch, sewn under prototype conditions. The rest of the size run is extrapolated from that single data point, and there are multiple places between that sample approval and the final garment where fit can degrade.

What you actually need is a pre-production size set: a sewn sample in every size, cut from the production fabric roll, before you commit to the full cut. This is where nearshore production has a real practical advantage. If your factory is four to six days away by sea rather than thirty, you can receive that size set, physically try the garments, measure them against your spec, and send back corrections before cutting begins. That feedback loop collapses from weeks to days. With a factory on a different continent, you're often making decisions from photos and PDF measurement reports, which is not the same thing.

Getting samples in seven to fourteen days — and getting a full pre-production size set back in your hands before the bulk cut starts — that's the difference between catching a grading error early and catching it after 200 units are cut wrong.

How Fabric Behaviour Between Rolls Causes Size Drift Mid-Run

Even if you've approved a fit sample and confirmed your grading, fabric can still move the numbers on you mid-run. And this happens more often than brands expect.

Different rolls of the same fabric from the same mill don't always behave identically. Weight, hand, and dimensional stability can vary between dye lots and between rolls within the same lot. In a 500-unit run, you might be cutting from six or eight rolls. If one roll has tighter tension from the beam, the garments cut from it will behave differently after washing than garments cut from a roll that was wound more loosely.

Shrinkage is the specific mechanism to understand here. Fabric shrinkage of 3% on a 60 cm chest measurement equals 1.8 cm of loss after first wash. In many jersey constructions, that pushes a size medium into the fit territory of a size small. And if different rolls in the same run are shrinking at slightly different rates, you end up with inconsistent results across units of nominally the same size. The garments look identical on the hanger. They don't fit the same on the body.

The fix is fabric testing, done on the actual production rolls before cutting — not on the sample yardage. Pre-washing test swatches from each roll, measuring shrinkage, and adjusting the cut pattern accordingly. Most factories know how to do this. Whether they do it on every run depends on whether the brand has explicitly required it in the production brief.

What a Proper Measurement Spec Sheet Needs to Include (and What Most Brands Leave Out)

A measurement spec sheet tells the production floor what the garment is supposed to measure. Most brands get the basics in there: chest, waist, hip, length. What they leave out is usually what causes the problems.

The most common missing element is the tolerance column. Measurement spec sheets submitted without tolerances are treated by most production floors as zero-tolerance documents. That sounds like a good thing, but it's functionally unworkable. Fabric is not sheet metal. You cannot cut to the millimetre every time. So when operators encounter a measurement that's 0.8 cm off and they don't have a tolerance to reference, they make a judgment call. They usually proceed. They don't document it. And across 500 units, those undocumented judgment calls accumulate.

What your spec sheet needs to include:

• Measurements for every size in the run, not just the base size
• A tolerance column for every measurement (typically ±0.5 cm to ±1.5 cm depending on the point of measure)
• How-to-measure instructions for non-obvious points, such as armhole depth or neck width
• The measurement state: relaxed flat, stretched flat, or on-body
• Whether measurements are taken before or after washing
• Grade rules between each size, not just the finished measurements at each end

That last point is important. If you only specify measurements for XS and XL, the factory has to interpolate the middle sizes. And interpolation means approximation. Specify grade rules, or specify every size explicitly. Either works. Leaving it to inference doesn't.

How Shrinkage Tolerance Interacts With Grading and Kills Fit at Scale

Grading and shrinkage are usually treated as separate technical concerns. They're not — they interact directly, and that interaction is where a lot of brands lose control of fit at scale.

Here's how it plays out. You've graded your size run correctly. You've built in the right increments. But you haven't accounted for the shrinkage differential between sizes. Larger sizes use more fabric area. More fabric area means more dimensional change under wash conditions. The shrinkage percentage might be the same across all sizes, but the absolute loss in centimetres is larger on bigger sizes. So a 3% shrinkage rate produces 1.5 cm of chest loss on a medium and 2.1 cm of loss on an extra-large. If your grading increments between those sizes were 2 cm, you've essentially erased most of the grade after first wash. The medium and XL fit nearly identically.

The solution is to build shrinkage compensation into the cut pattern, not the washed measurement. This means cutting larger than the finished spec by a factor that accounts for expected shrinkage, verified by testing the actual production roll. It's standard practice. But it requires a shrinkage test result before cutting begins, and as noted, factories working fast don't always wait for it.

What In-Line Inspection Checkpoints Catch Before Hundreds of Units Go Wrong

In-line inspection means checking the work at specific stages during production, not just auditing finished garments at the end. For fit and sizing, the critical checkpoints are at the cut stage and at the first-off-the-line stage after sewing begins.

At the cut stage, colour-coded cut bundles by size are the simplest and most effective control. Each size gets a distinct colour tag attached to the bundle before it goes to sewing. Bundle counts are verified against the cut plan. Pieces are checked against the spec before the bundle is released. This costs almost nothing to implement. It's absent from a significant proportion of small-run factory workflows because nobody asked for it. Ask for it.

At the first-off-the-line stage, the first completed garment from each size is measured before the rest of the units in that size proceed through sewing. If a systematic error is present in the cut or in the sewing construction, this is where you catch it. In a 500-unit run across four sizes, catching a grading error after cutting but before sewing can still result in re-cut costs on 30 to 40% of the run and add 8 to 12 days to delivery. Catching it before sewing, via a proper first-off inspection, is significantly less damaging.

These checkpoints aren't complicated. They're a list of measurements, a person with a tape measure, and a record of what was found. The value is in making them non-negotiable and documenting the results. If a factory can't tell you what their in-line inspection process is, that's worth discussing before your order goes into production.

The Return Rate Maths: What One Centimetre of Chest Discrepancy Costs You in Ecommerce

Let's put numbers to this. A U.S. boutique brand ships 300 units of a dress style at an average retail price of $145. They experience a 28% return rate driven by fit inconsistency — which, at that price point and that category, is a realistic figure, not a worst case. That's 84 returns. Factor in reverse logistics, restocking labour, quality inspection on returned units, and the margin lost on product that can't be resold at full price. You're absorbing roughly $12,000 on that single SKU. That's not a marketing budget problem. That's a production spec problem.

And the damage compounds. Ecommerce platforms measure return rates by seller. High return rates affect ranking, advertising costs, and in some cases seller status. Customers who return don't usually come back. The $12,000 you can quantify is probably not the full cost.

Now consider what it would have cost to prevent it. A pre-production size set in every size: a few hundred dollars and a week of time. A shrinkage test on each production roll: a day. A measurement spec sheet with tolerances properly filled in: an afternoon of work. In-line bundle checks during cutting: essentially zero direct cost. The prevention budget for a problem that costs $12,000 downstream is almost comically small by comparison.

The short version of all of this: sizing consistency is not something you achieve once with a good fit sample and then maintain passively. It requires explicit written specs with tolerances, pre-production size set approval on the actual production fabric, fabric shrinkage testing per roll, and in-line checks during the cut and the first units off the sewing line. None of that is exotic. It's just methodical. The brands that have low return rates are the ones who treat every one of those steps as non-negotiable, not optional.