Yes. The strongest evidence points to reducing uncertainty before checkout, rather than simply adding more sizing information. Fit/style accounts for roughly 70% of apparel returns in McKinsey’s survey, and size/fit was the leading reason in Coresight’s apparel survey.
Tactics that have actually moved return rates
-
Personalized fit recommendations based on purchase/return history
- Don't just ask for height/weight. Use what the shopper previously bought, kept, returned, and how those garments fit.
- This is particularly powerful when recommendations are SKU- or cut-specific rather than saying "you normally wear M."
- Under Armour Japan, for example, reported a 27% YoY reduction in size-based returns after integrating personalized fit recommendations.
-
Show the garment on bodies that resemble the shopper
- Multiple models with different heights, sizes, body shapes and skin tones can reduce the "model looks great, but will it look like that on me?" gap.
- McKinsey specifically identifies more representative model imagery as a return-reduction tactic.
- I'd prioritize real customer photos/reviews wearing the exact SKU over generic UGC.
-
Virtual try-on / visual fit
- This goes beyond a size chart by answering the more important question: What will this actually look like on me?
- A randomized field experiment found that virtual fit information increased conversion/order value while reducing returns and "home try-on" behavior such as ordering multiple sizes.
- A TA3 Swim case study reported a 47% lower size-related return rate after implementing mobile body measurement and fit recommendations. That's vendor-reported, so I'd treat the magnitude cautiously, but the direction is compelling.
-
Attack "bracketing" specifically
- If customers routinely buy S+M+L and return two, your problem isn't just sizing—it is using your fulfillment network as a fitting room.
- Track "same SKU, multiple sizes in one order" as its own KPI.
- Then test interventions such as a confident size recommendation, "most likely to fit" messaging, or showing a comparison between the shopper's previously purchased size and the current garment.
- Bracketing is a substantial return driver in online fashion.
-
Make fit language much more granular
Instead of:
"Runs true to size."
Give shoppers something like:
- Shoulder: fitted
- Bust: relaxed
- Waist: fitted
- Length: runs short
- Fabric stretch: high
- Recommended if between sizes: size up
This is particularly useful because "size" and "fit" aren't the same thing. A customer can buy the correct numerical size and still dislike the silhouette.
-
Feed return reasons directly back into merchandising/product development
- Don't treat "too short," "tight in shoulders," "sheer," "color different," and "fabric thinner than expected" as customer-service data.
- Aggregate them by SKU × size × color × supplier × season.
- If one style generates disproportionate returns for "tight arms," fix the pattern or grading rather than trying to educate the next customer.
- This is one of the less glamorous but potentially highest-ROI interventions.
-
Improve expectation-setting, not just fit
Returns aren't exclusively sizing problems. Coresight found color and damage among the other major return reasons.
Useful tactics include:
- video showing fabric movement/drape
- close-ups showing texture and opacity
- natural-light color photography
- explicit "sheerness" and "stretch" ratings
- garment measurements in addition to body measurements
- photos of the exact colorway rather than one generic product image
-
Use returns themselves to personalize future shopping
If Jane bought a medium, returned it as "too tight in bust," and kept the large, the next product shouldn't treat Jane as a generic M/L shopper. Her return is valuable fit data.
One interesting academic finding is that previous purchases of the same SKU or SKU-size combination reduced return likelihood—suggesting that behavioral history can be more useful than a generic size finder.
One important caveat
I'd be careful about buying a "size recommender" and assuming returns will automatically fall. A 2025 study using 496,365 apparel orders found that simply using a size finder was actually associated with a slightly higher return probability; the authors found that the shopper's previous purchase behavior was more predictive.
So the hierarchy I'd use is:
1. Fix inconsistent product fit → 2. Communicate actual garment behavior → 3. Personalize using customer history → 4. Show representative bodies/UGC → 5. Add virtual try-on where it makes economic sense → 6. Target bracketing.
And I'd measure return rate by reason, not just overall return rate. Otherwise you can spend heavily on fit technology while your real problem is "color looked different" or "fabric wasn't what I expected."
If you're running an apparel brand, I can also lay out a 90-day return-reduction experiment roadmap ranked by expected ROI and implementation difficulty.
