Every return you make is also a piece of research. When you tick "doesn't fit" or "not as described" and drop a parcel at the post office, that reason code travels back to a brand's data team and sits alongside thousands of others. What happens next is less visible than the refund hitting your account—but arguably more consequential for the clothes you'll see next season.
Key takeaways
- Return reason data is one of the most direct feedback channels brands have, because it captures a decision made at the moment a garment failed.
- Sizing and fit are consistently the most common return drivers in online fashion, which is why they attract the most analytical attention.
- The same data that helps brands improve products also informs decisions about which styles to discontinue, which fabrics to replace, and how to rewrite a product description.
- Tools that map body measurements to size recommendations are increasingly used to close the loop between what a brand makes and what a customer actually needs.
- Not all of this work benefits the shopper equally—some brands use returns intelligence to tighten return policies rather than improve products.
Why does this matter to you as a shopper?
Returns are enormous in scale. According to the National Retail Federation and Happy Returns, total retail returns in the US were projected to reach $890 billion in 2024, with retailers estimating that roughly 16.9% of annual sales would come back. In European online fashion specifically, the European Environment Agency estimates that the average return rate for clothing bought online sits at around 20%—one in every five garments—and that on average a third of those returned items end up destroyed rather than resold.
Those numbers represent real cost, real waste, and real information. Brands that treat returns as data rather than simply as logistics problems are doing something genuinely useful—even if the motivation is commercial.
How do brands actually analyse return data?
Most brands collect return reasons at the point of return: a dropdown on the returns portal, a tick-box on the slip inside the parcel, or a short survey. These reasons are then aggregated by product, by size, by region, and by time period. The patterns that emerge are compared against sales data, customer reviews, and in some cases body measurement data from fit tools. What follows are the six most common ways that analysis feeds back into product decisions.
1. Adjusting size grading between seasons
When a disproportionate share of returns on a specific style come back with the reason "too small" or "too large," and that pattern holds across multiple sizes, it usually points to a grading problem rather than individual customer preference. Grading is the process of scaling a base pattern up and down across a size range, and small errors compound: a base pattern that runs slightly narrow will produce a size 14 that runs narrow too.
Brands use return rate by size as a diagnostic. If size 10 in a particular trouser style returns at twice the rate of size 12, the pattern team will pull the measurements and look for where the grade stepped incorrectly. The fix might be as small as adding two centimetres to the hip grade for the next production run—invisible to most shoppers, but meaningful to the ones who kept returning that size.
This kind of correction is painstaking to do manually. Platforms like True Fit, which aggregates fit and preference signals from a large base of shoppers, can surface these discrepancies at scale, flagging which products are generating fit-related returns before a brand's internal team has had time to notice the pattern.
2. Rewriting product descriptions to set accurate expectations
A return reason that reads "not as described" or "colour different from photo" is a content problem, not a product problem. The garment may be exactly what the design team intended—but if the product page described it as "relaxed fit" when it actually cuts close to the body, or showed it in a lighting condition that made navy look black, the return was predictable.
Brands track which products generate high rates of expectation-mismatch returns and audit the copy and imagery. Common fixes include adding a "fit note" to the product description ("this style runs slim through the thigh—size up if you prefer room to move"), shooting the garment on a wider range of body types, or adding a fabric weight and drape description so shoppers can judge for themselves.
This is one of the cheapest interventions available to a brand, and the data makes it easy to prioritise: the styles with the highest "not as described" return rates are the ones that need the most urgent copy review.
3. Identifying fabric and construction failures
Some return reasons point directly at the physical object: "pilling after one wash," "seam came apart," "fabric thinner than expected." When these cluster around a specific fabric or supplier, they become quality intelligence.
Brands use this data to renegotiate fabric specifications with mills, to add wash tests to their quality control process for specific materials, or in some cases to drop a fabric entirely from future ranges. A jersey that returns at a high rate for pilling in its first season is unlikely to appear in the next one—or if it does, it will be from a different supplier with a tighter specification.
This feedback loop is slower than the sizing corrections above, because it requires enough returns to accumulate to be statistically meaningful, and because changing a fabric mid-development cycle is expensive. But over multiple seasons, return data becomes one of the most reliable inputs into a brand's material library decisions.
4. Informing size curve planning
A size curve is the distribution of units a buyer orders across the size range—how many XS, how many S, how many M, and so on. Getting this wrong is expensive: too many units in a size that doesn't sell means markdowns; too few in a size that sells out means lost revenue and frustrated customers.
Return data feeds into size curve planning in a less obvious way. If a brand notices that size XS has a very low return rate and size M has a high one, it may indicate that the XS customer is getting exactly what she expected while the M customer is not—which might mean the M grade is off, or that the product is being bought by a customer whose body shape doesn't match the brand's block. Either way, the next buy might shift units toward the sizes that are performing well.
Merchandising tools like Style Arcade bring together sales, stock, and returns data into a single planning view, helping buying teams make these adjustments with more precision than spreadsheets allow.
5. Refining fit recommendations before the customer even buys
Some brands have moved the intervention upstream: rather than waiting for a return to happen and then analysing it, they use body data and purchase history to recommend the right size at the point of purchase. The logic is that a return that never happens is better than a return that teaches you something.
This is where platforms like Bold Metrics, which builds digital body profiles from a shopper's measurements, and True Fit, which draws on a large network of size and preference data, come in. Both are used by brands to surface size recommendations on product pages—and both feed the resulting purchase and return outcomes back into their models to improve future recommendations.
Zalando, which operates one of Europe's largest fashion platforms and connects tens of millions of active customers with thousands of brands across its markets, has invested significantly in this space. Its research team published work on a probabilistic model called SizeFlags, designed to reduce size-related returns across multiple countries by learning from weakly annotated customer data at scale. The approach treats the return signal not as an outcome to be managed after the fact, but as a training input for a model that gets better at predicting fit over time.
6. Deciding which styles to discontinue
Not every return is a fixable problem. Sometimes a style returns at a persistently high rate across multiple seasons, across multiple sizes, and with reasons that don't point to a single correctable issue—"just didn't like it," "not flattering," "changed my mind." When that pattern holds, it usually means the product doesn't resonate with the customer it was designed for, regardless of how well it's made.
Brands use return rate as one input into discontinuation decisions. A style that sells moderately but returns at a high rate may actually be generating less net revenue than its sell-through figures suggest, once the cost of processing returns is factored in. Return rate per style, when tracked consistently, becomes a proxy for genuine customer satisfaction—distinct from the purchase decision, which can be influenced by a good product photo or a promotion.
This is one of the more commercially significant uses of returns data, and one of the least visible to shoppers. The style that quietly disappears from a brand's range after one season may have been discontinued not because it didn't sell, but because too many of the people who bought it sent it back.
What does this mean for you?
The feedback loop your returns create is real. A reason code you select in thirty seconds on a returns portal can, in aggregate with thousands of others, change the grading of a pattern, prompt a rewrite of a product description, or retire a fabric from a brand's range. That's a form of consumer influence that operates quietly and at scale.
It also means that being specific when you return something is worth doing. "Didn't fit" is less useful to a brand's data team than "too narrow across the shoulders" or "runs large in the waist." The more precise the reason, the more actionable the signal—and the more likely the next version of that garment is one that works.
The less comfortable side of this picture is that returns intelligence is also used to tighten return windows, flag accounts with high return rates, or charge return fees. The data flows in both directions: it can make products better, but it can also be used to make returning harder. Knowing that your returns are being read is useful context for deciding how and when to use them.
FAQ
Does returning clothes actually change how brands design products? Yes, in aggregate. Return reason data is one of the most direct signals brands receive about fit, fabric, and expectation mismatch. Individual returns matter less than patterns across thousands of customers, but those patterns do feed into grading, material, and copy decisions.
What return reason is most useful for a brand to receive? Specific fit feedback—"too narrow across the shoulders," "runs large in the waist"—is more actionable than a generic "didn't fit." Brands can map specific fit complaints to pattern measurements; generic reasons are harder to act on.
Do brands use my body measurements to reduce returns? Some do. Platforms that build size recommendations from body data, such as Bold Metrics and True Fit, help brands suggest the right size before a purchase is made. The goal is to reduce returns before they happen rather than only learning from them after.
Is my returns data shared with other brands? This depends on the platform. Some fit recommendation tools operate across multiple brands and use aggregated, anonymised data to improve recommendations network-wide. Individual brands' own returns portals typically keep data within that brand's systems.
Why do some styles disappear after one season? High return rates are one factor in discontinuation decisions. A style that sells but returns frequently may generate less actual revenue than its sales figures suggest, and brands track return rate per style as a measure of genuine customer satisfaction.
Further reading
- The destruction of returned and unsold textiles in Europe's circular economy — European Environment Agency
- NRF and Happy Returns: 2024 Retail Returns Report
- SizeFlags: Reducing Size and Fit Related Returns in Fashion E-Commerce (KDD '21)