Online apparel returns reached an average of 24.4% for the 12 months ending March 6, 2023, in a Coresight Research survey of U.S. apparel brands and retailers, and size or fit was cited by 53% of respondents as the leading reason (Coresight Research). That reframes the problem online shopping creates. Returns aren't merely a warehouse or delivery issue. They begin earlier, on a product detail page where shoppers must decide whether a garment will fit, drape, and look as expected without touching it or trying it on.
For apparel retailers, the critical question isn't only how to process unwanted items more efficiently. It's how to give shoppers enough confidence to place the right order the first time. Static charts, inconsistent measurements, generic imagery, and weak product descriptions leave customers managing uncertainty through abandonment or bracketing, ordering several sizes and returning the extras.
Table of Contents
- The Scale of the Online Shopping Problem in Apparel
- Where Shoppers Lose Confidence on the Product Page
- Wrong Size Versus Wrong Expectation
- How Bracketing Inflates Hidden Costs
- Solutions That Close the Confidence Gap
- Prioritizing Fixes for Maximum Impact
The Scale of the Online Shopping Problem in Apparel
Reverse logistics turns an uncertain product decision into an operating cost. Transportation, inspection, repackaging, and inventory handling all consume resources after an item comes back. Returned merchandise can also lose selling momentum when it misses a campaign window, arrives after seasonal demand peaks, or requires a markdown before resale. The retailer absorbs reverse-logistics costs that can exceed the original shipping revenue on each returned item.
The shopper carries a different cost: uncertainty, inconvenience, and the possibility that the preferred item will no longer be available in the right size. That cost begins before checkout. A product page that leaves fit or appearance unclear can produce either abandonment or a defensive order containing multiple sizes.
The information gap behind the return
Physical retail lets customers inspect fabric, compare proportions, and test fit before paying. Online shopping removes those signals while leaving sizing fragmented across brands, regions, and product types. A medium in one catalog may not correspond to a medium in another, and a numerical size cannot fully describe how a garment will sit on a particular body.
The issue extends beyond the size selector. A single model, pose, or camera angle forces shoppers to infer drape, silhouette, and coverage. Static measurements may also leave unclear whether they describe the body or the finished garment. These gaps make the product detail page a confidence test, not merely a catalog entry.
A return-rate problem that starts with missing product information will not be solved by reverse logistics alone. Better return processing limits operational waste, but it does not repair the purchase decision that created the return.
Retailers need fit and appearance information before checkout, then measurement practices that connect product-page behavior with profitable outcomes. That includes ecommerce marketing attribution, which helps teams distinguish traffic that converts profitably from traffic that produces costly, uncertain orders. The practical priority is to reduce uncertainty at the product page, where shoppers decide whether to buy one item confidently, abandon it, or bracket the purchase with several alternatives.
Where Shoppers Lose Confidence on the Product Page
Roughly 40% of shoppers abandon apparel purchases because of the product-detail-page experience, according to a 2025 industry report (Coresight Research activity). That figure places fit anxiety before the cart, not after delivery. A shopper can like the product, accept the price, and still leave when the page cannot support a confident size decision.
The failure often begins at the size selector. A chart may list body measurements without clarifying whether they refer to the body or the finished garment. The model's height and selected size provide context, yet they cannot show how the same dress will sit on different proportions. One pose and camera angle leave shoppers estimating silhouette, coverage, and drape.
The journey then follows a predictable chain: incomplete evidence creates uncertainty, uncertainty delays the size choice, and delayed choice increases the chance of exit or a multi-item order. The retailer sees a product-page departure, while the shopper may be managing perceived fit risk before any return exists.
A static chart makes customers translate measurements without personal interpretation. Generic imagery shows one person's result, not the visitor's likely fit. Limited viewpoints can conceal length, stretch, rise, or shape. Inconsistent descriptions force shoppers to reconcile product data themselves, while “true to size” provides little help unless the reference standard is clear.
A practical guida CRO per PMI can help teams audit page friction, but a general conversion review may miss the apparel-specific risk. A confusing button is easy to identify. A size selector that asks customers to make a high-consequence decision with weak evidence requires product, fit, and behavior data together.
Review the path from product entry to size interaction, then compare it with product-level return comments. Examine exits near size selection, gallery engagement, and recurring questions about fit or appearance. Retailers can use this guide to whether a garment fits to test whether their own PDPs answer the questions shoppers ask before ordering.
The following video adds context on the online fit experience. It should support, not replace, evidence specific to each product page.
Wrong Size Versus Wrong Expectation
A garment can disappoint for two different reasons. It can be the wrong size, or it can be the right nominal size but fail to match what the shopper thought they were buying.
Wrong size means the physical relationship between the garment and the body is incorrect. The waistband may be tight, sleeves may be short, or the shoulder line may sit differently than expected. The useful remedy is better measurement interpretation, garment-specific sizing, and fit guidance that accounts for body shape rather than presenting a universal label.
Wrong expectation is different. The shopper may receive an item that technically matches the ordered size but looks unlike the product-page impression. Color, fabric behavior, silhouette, opacity, length, and drape can all create that mismatch. A size recommender can't fully repair an image that sets an inaccurate expectation.
Two return diagnoses require two interventions
Independent summaries identify sizing, fit, and color as accounting for 45% of retail returns, while products that don't match their descriptions or images remain a consistent secondary reason (SizeMarker). The figure is useful because it separates the physical measurement problem from the visual communication problem.
| Failure on the page | What the shopper lacks | Appropriate intervention |
|---|---|---|
| Fit uncertainty | Confidence that the selected size will work | Personalized size recommendation |
| Silhouette uncertainty | A sense of how the garment sits on the body | Virtual try-on and additional views |
| Color uncertainty | Confidence that the displayed shade reflects the product | Better calibrated imagery and clear descriptions |
| Construction uncertainty | Evidence about stretch, weight, and finishing | Material details, close-ups, and movement views |
The distinction changes how a retailer reads return data. A high volume of “too small” comments points toward measurements, grading, or recommendation quality. Comments about color, shape, or appearance point toward imagery and expectation management. Mixed feedback means the PDP may have both problems, and a single “add a size chart” response won't address them.
The strongest pages make uncertainty visible and answer it directly. They show the garment from useful angles, explain how it fits, and let shoppers evaluate the product against their own context. That approach doesn't promise perfect prediction. It reduces the amount of guesswork required to make a purchase.
How Bracketing Inflates Hidden Costs
Bracketing is the shopper's rational response to missing fit information. A customer orders multiple sizes, keeps the one that works, and sends the others back. A peer-reviewed study of online apparel purchasing describes this behavior as a way to manage uncertainty when fit can't be verified before purchase and return shipping is free (peer-reviewed bracketing study).
From the shopper's perspective, the approach lowers the risk of ending up with nothing. From the retailer's perspective, it creates a temporary demand signal that overstates the number of units likely to remain sold. Several units leave the warehouse for one eventual purchase, then the extras re-enter a process that requires transport, inspection, restocking, and sometimes repricing.
The inventory effect is easy to miss
Bracketing doesn't only increase the number of parcels moving through the network. It complicates forecasting. A retailer sees demand for multiple sizes, allocates inventory to orders that may not become retained sales, and waits for returned units before knowing what stock is available.
That uncertainty can affect customer service in practical ways:
- Availability becomes less reliable: Units in transit or return processing aren't immediately sellable.
- Replenishment signals become noisy: Orders represent trial demand, not necessarily retained demand.
- Warehouse labor rises: Teams must inspect and route extra units instead of fulfilling new orders.
- Markdown exposure increases: Delayed stock may miss the period when the original price was strongest.
The study's central implication is more important than the behavior itself. Bracketing isn't evidence that shoppers are careless. It's evidence that the retailer hasn't supplied enough information for a confident single-size decision.
Operational rule: Treat bracketing as a symptom of information asymmetry, not as a customer-behavior problem to punish.
Return-policy restrictions may reduce some bracketing, but they can also increase purchase hesitation when shoppers still lack fit confidence. Pre-purchase guidance attacks the cause instead of shifting more risk onto the customer. A product page that recommends a garment-specific size, explains the basis for that recommendation, and shows the silhouette on a relevant body gives shoppers a reason to order one considered option rather than several guesses.
Solutions That Close the Confidence Gap
The highest-impact changes operate on the product detail page, where shoppers choose a size and form an expectation of the garment. Guidance should remain in that decision area, without forcing customers to calculate measurements, leave the page, or interpret a generic chart alone.
Start with fit-specific guidance
An AI size recommendation can collect inputs such as height, weight, age, and body shape, then convert them into a garment-specific suggestion. The label is only useful when the system connects it to the item's own size chart and displays the result beside the size selector.
A 2021 SizeFlags A/B test found that size advice reduced returns by 4.3% for items flagged as too small and 6.6% for items flagged as too big (SizeFlags paper). The finding points to a practical product-page intervention: better guidance can change the decision before checkout, not merely explain a return afterward.
Effective implementation should account for differences in how shoppers provide information:
- Low-friction entry: Use a short questionnaire instead of requiring manual garment measurement.
- Optional visualization: Offer a selfie-based route and a model-selection alternative for shoppers who do not want to upload an image.
- Garment-level output: Recommend a size for the specific item, rather than displaying only a general account preference.
- Unit flexibility: Accept metric and imperial inputs so the tool does not create another translation task.
- Clear explanation: Explain why the recommendation appears without exposing unnecessary personal details.
Add visual expectation management
Sizing guidance addresses measurement uncertainty. Virtual try-on addresses the separate question of how the garment will look, including silhouette, proportion, and appearance on a shopper-specific body model. Advanced virtual fitting-room research reported an average 36.5% reduction in returns alongside conversion gains (virtual clothing try-on overview). For categories where silhouette and fabric behavior materially affect fit decisions, this approach shows measurable return reductions.
Robosize is one platform retailers can evaluate. It generates a shopper-specific body model from a short questionnaire and optional selfie, renders apparel on that model, and displays a per-garment size recommendation on the product page. Its documented implementation paths include a one-click Shopify app and a JavaScript snippet for other ecommerce platforms.
The objective is decision evidence, not extra decoration. Multi-angle views, fit output, and product-specific context should appear before the shopper commits to an order. Together, these elements reduce the gap between what the PDP promises and what the customer expects to receive.
Prioritizing Fixes for Maximum Impact
Retailers can reduce the pre-purchase confidence gap without rebuilding the entire storefront. The priority is to sequence changes by uncertainty, traffic, and their effect on sizing decisions, visual expectations, abandonment, and bracketing.
Begin with product pages that attract substantial traffic and generate repeated fit-related returns. Narrow the first rollout to high-volume categories such as denim, dresses, activewear, or footwear when the data supports it. A focused test makes shopper behavior, recommendation use, retained purchases, and return reasons easier to compare across similar products.
A practical order of operations
Audit the decision failure. Separate size-related returns from complaints about appearance, color, or product descriptions. Review exits and engagement around image interaction, size selection, and add-to-cart behavior.
Repair the information layer. Standardize product measurements, size-chart mapping, fit descriptors, fabric details, and imagery within the selected category. A sizing tool cannot correct inaccurate or inconsistent source data.
Add personalized sizing. Place the recommendation beside the size selector, where the choice occurs. Keep the questionnaire short and explain the result in terms shoppers can understand.
Test visual try-on selectively. Apply it to products where silhouette, proportion, or drape materially affects the decision. Compare interaction, conversion, and return outcomes with comparable product pages instead of applying the same treatment across the catalog.
Measure retained value. Review conversion, average order value, return rate, exchange behavior, recommendation adoption, and inventory availability together. A conversion increase is not a successful test if it produces more unprofitable returns.
Vendor evaluation belongs in the implementation plan. Check product-to-size-chart matching, metric and imperial unit support, mobile performance, privacy controls, analytics, platform integration, and usage limits. A tool can appear effective in a demonstration yet underperform when its recommendation loads late, displays poorly, or asks shoppers to repeat information.
Use this guide on reducing ecommerce returns as an operational checklist, then connect its recommendations to category-level return reasons and PDP analytics. The highest-impact intervention is the one that removes a major source of uncertainty without adding comparable effort or friction.
Retailers who audit PDPs against the uncertainty types identified earlier and address the largest measurable gap first are positioned to reduce both bracketing and abandonment faster than retailers applying broad, undifferentiated changes. Robosize provides AI size recommendations and shopper-specific virtual try-on through a Shopify app or JavaScript integration for other stores. Retailers can visit Robosize to evaluate whether a fit and visualization layer suits their highest-return categories.