Ecommerce returns are online orders that customers send back for a refund, an exchange or store credit. They are common: US retailers expected 19.3% of online sales to come back in 2025, against 15.8% of all retail sales, according to the National Retail Federation. The returns a store can actually prevent mostly start before the order is placed: the product arrives and does not match what the page said or showed. The size runs small, the drill comes without the battery the shopper assumed was included, the oak looks orange instead of pale, the "navy" sweater is closer to black.
Damage in transit and changed minds are real too, but those are logistics and policy problems. Size, specs, photos and color are product data problems, and you fix them field by field. Below you find which return reasons a product page can prevent, a fix list per reason with the exact fields involved, how the EU right of withdrawal changes the stakes, and how to find the products that cause most of your returns.
How many online orders come back, and what shoppers say about why
Return rates differ a lot by category, so treat any single number as a reference point rather than a target:
- The 19.3% comes from the National Retail Federation and Happy Returns, based on surveys of 2,006 US consumers and 358 ecommerce professionals at large retailers (NRF, October 2025).
- Apparel is higher. Coresight Research estimates the US online apparel return rate at 23.4% in 2025, barely down from 24.4% in 2023. In its February to March 2026 survey of 212 US shoppers who had returned apparel bought online, around 70% named size or fit as a reason. The report was sponsored by Alvanon, a sizing technology company.
- Zalando reports that about a third of its returns are size-related, and that size advice reduced size-related returns by 10% compared with items without it (Zalando).
- Across categories, 45% of shoppers in the US, Canada and the UK said they had returned an online purchase because of incorrect or misleading information, in a survey of 2,712 shoppers that the PIM vendor Salsify ran in October 2025 and published in January 2026.
Why US shoppers returned apparel bought online
| Category | Share of respondents |
|---|---|
| Size or fit | 69.8% |
| Material quality | 27.8% |
| Color | 24.5% |
| Delivery time | 20.3% |
| Damage or defect | 17.5% |
| Found cheaper elsewhere | 17% |
| Overcharge | 11% |
These surveys ask different people different questions, so do not add them up. The pattern they share is what matters: size and fit lead in fashion, and mismatched information is a reason in every category. To calculate your own rate per product or category, the return rate calculator does the arithmetic.
Which return reasons a product page can prevent
Before fixing anything, sort your reasons. A useful reason list is short and fixed, so customers and staff pick from it instead of typing free text. The nine reasons below are a practical starting set; the third column tells you where to work.
Return reasons and who can prevent them
| Too small / too large | The size label did not match my body or my other clothes | Yes | Garment measurements, fit, model size, size system |
|---|---|---|---|
| Not as described | It is a different material, size, capacity or version | Yes | Attributes, title, description, what's in the box |
| Looks different from the photos | Quality, scale or finish differ from the pictures | Yes | Images per variant, scale shot, detail shots |
| Color differs | The color is not what I expected | Yes | Images per color, color family, color name |
| Does not fit my device | Wrong model, connector, plug or bulb fitting | Mostly | Compatibility list, plug type, fitting, model numbers |
| Ordered several sizes | I kept one and sent the rest back | Partly | Size chart and fit notes that reduce doubt |
| Changed my mind | Found it cheaper, no longer needed | Rarely | Price consistency across channels |
| Damaged or defective | Broken in transit or faulty | No | Packaging, quality control, carrier |
| Wrong item received | The warehouse picked another SKU | No | Picking process, unless two SKUs share a barcode |
Note the last row. A "wrong item" return is usually a picking error, but it can also be a data error: two products sharing one EAN or a supplier SKU mapped to the wrong product will send the wrong box every time. Our post on dirty data shows how to find duplicate barcodes and other errors of that kind.
Size and fit: what a size chart needs to answer
A size chart that only lists S, M and L next to chest circumference is not enough. Baymard Institute found that 83% of desktop and 87% of mobile apparel sites did not provide sufficient sizing information in its 2022 testing. Its list of what a good size guide contains includes numeric sizes, measurements in both centimeters and inches, international conversions, instructions for measuring yourself, details specific to the product type, a link next to the size selector and the model's measurements with the size they wear.
Here is what that looks like as data for a men's merino crew-neck sweater in size M, sold in the EU and the UK:
- size_label: M, with size_system EU and a UK conversion stored per size, not typed into the description
- chest_width_flat_cm: 54, body_length_cm: 70, sleeve_length_cm: 64, measured on the garment, per size
- fit: regular (from a fixed list: slim, regular, relaxed, oversized)
- stretch: low, because merino jersey gives less than a cotton blend
- model_height_cm: 186, model_size_worn: M
- fit_note: a short line you update from return data, for example "runs half a size small in the shoulders"
Footwear needs the same treatment: insole length in centimeters per size, width, and a fit note per last. Furniture and home goods have their own version of fit: the sofa that does not go through the door is a size return too. Store outer dimensions, packed dimensions and the smallest opening the product must pass through as separate numeric fields.
Spec mismatches: when the product does something the page did not say
"Not as described" returns are rarely about outright false claims. They come from gaps and contradictions: the title says "18V cordless drill" and the photo shows a battery, but the product is sold body only; a phone case listed for a phone model when it only fits the larger Plus version; a pendant lamp with an E27 fitting bought for an E14 bulb; a duvet cover in "double" size that follows UK dimensions in a Dutch store. Each could have been prevented by one structured field shown on the page.
- What's in the box: a list field, not a sentence buried in the description. "Battery not included" belongs in its own attribute and, for tools, in the title.
- Compatibility: model numbers, connectors, fittings and plug types as attribute values that filters can use, so a shopper can rule a product out before buying.
- Dimensions and capacity: numbers with units in separate fields (40 x 60 cm, 1.5 l, 2.0 Ah), consistent across title, attributes and description.
- Material and composition: percentages for textiles, wood species and finish for furniture, coating for cookware.
- One source of truth: when the title says 4.0 Ah and the attribute says 2.0 Ah, one of them is wrong and the shopper will find out which. Generate titles and descriptions from the attributes, not the other way around.
Photos and color: closing the gap between the picture and the parcel
Shoppers look before they read. In Baymard's testing, 56% of users started exploring a product page with the images. When the images mislead, the text rarely corrects the impression. Typical causes of "looks different" and "wrong color" returns:
- One image set for every color. The black variant's photos appear on the sand and olive variants. Attach images per variant, and make the first image of each variant show that color.
- Marketing color names only. "Storm" tells the shopper nothing. Store a standard color family (gray, blue) next to the brand's color name, so both the filter and the page are clear.
- No sense of scale. A 25 cm vase photographed alone on white could be 10 or 50 cm. Add one photo in context or in hand, and show the main dimensions near the images.
- Supplier images of another version. Old packaging, a US plug, the previous model year. Check supplier images against the product you actually stock.
- Heavy retouching. Color correction that makes untreated oak look like honey. Shoot under neutral light and compare a print with the real product.
Google Merchant Center already asks for a main image that shows the exact item being sold, so fixing this also keeps your Shopping listings compliant. Our guide to product image quality covers resolution, angles and naming.
Answer the questions shoppers ask before they order
Some returns start with a question nobody answered. Will this charger work with my phone? Is the sweater warm enough without a coat? Does the folded stroller fit in a car trunk? When the product page stays silent, the shopper guesses and orders, then sends the product back when the guess was wrong, or orders two sizes and keeps one. A short FAQ on the product page catches these doubts where the buying decision happens.
Useful product FAQs are specific to the product and come from your data: fit advice from your size and fit attributes, compatibility from your model lists, care from your material data. Store-wide questions about shipping and the returns policy belong on a help page or once per category, not copied by hand onto every product. Your return comments are the best source of questions: every recurring "I expected..." is a question the page should have answered.
WISEPIM's Product FAQs work this way. You write questions in a product's FAQs tab, or let WISEPIM AI draft them from the product's fields, attributes and web research, using question types picked for your industry, such as Sizing and fit for fashion or Compatibility for electronics. Each AI answer cites what it was written from, and a question your data cannot back is skipped rather than guessed. Drafts can wait for your review before they go live. Shared answers, such as warranty or returns, go on a category once. On export, WISEPIM adds the FAQ block with FAQPage schema to the product description and sends the questions as structured data to Shopify, WooCommerce and Magento 2.
The EU 14-day withdrawal right and why accurate data limits its cost
In the EU, online shoppers can return goods without giving a reason. Under the Consumer Rights Directive, the consumer has 14 days to withdraw from a distance contract, counted for goods from the day they receive them. The trader must refund the payment, including standard delivery costs, within 14 days of being told. The consumer pays only the direct cost of sending the goods back, and only if the trader said so beforehand, and is liable for loss of value only when the goods were handled beyond what is needed to establish their nature, characteristics and functioning. The same directive requires the trader to give the main characteristics of the goods before the order.
A product that does not match its description is a different case. Under the Sale of Goods Directive, goods must match the description, type, quantity and quality agreed in the sales contract, and the seller is liable for a lack of conformity that becomes apparent within two years of delivery; member states may set longer periods. A "not as described" return can therefore fall outside the 14-day window and the return-cost rules you set for withdrawals. This is general context, not legal advice; national laws add details.
The practical point: every attribute on your product page is part of what you promised. Accurate, complete product data reduces returns and also limits the returns where you carry the full cost.
Product data fixes for each preventable return reason
Work through this list per category, starting with the categories that have the highest return rate and the highest share of preventable reasons:
Return-reducing fixes in your product data
Tracing returns back to the products and fields behind them
Store-wide return rates hide the problem. Most of the preventable returns usually come from a small group of products, so calculate the return rate per SKU and per variant, and compare it with the average of its category. Wait for enough orders before you judge: one return on five orders is 20%, but it is not a pattern yet. Then read the reasons and comments for the products at the top of the list and check which fields on their pages are empty, vague or contradictory.
The quality of this analysis depends on how returns are recorded. Three habits make the data usable. Keep "too small" and "too large" as separate reasons, because the direction tells you what the fit note should say. Record the return against the variant (size and color), not only the parent product, since one size often causes most of the trouble. And keep a free-text comment next to the reason list: the comment is where shoppers write "the sleeves are short" or "no charger in the box", which is exactly the field you need to fix. Let the warehouse add its own note when the item comes back, so you can tell a fit problem from a damaged parcel.
WISEPIM's returns analytics does this matching with order data from connected Shopify, WooCommerce, Magento 2, CCV Shop and Lightspeed stores. It shows the return rate per product next to your store's overall rate and groups returns by reason where the store platform records one. Field Health in the Data Quality report then shows which fields are missing on how many products, so you can see what to fill first.
We wrote earlier about how accurate descriptions cut the waste that returns create, in Stop e-commerce waste. The fixes in this post are the field-level version of that idea: measure returns per product, read the reasons, and fill or correct the fields that would have answered the shopper's question before they ordered.
