Ecommerce Returns in 2026: How to Find Out Why Items Come Back

What Are Ecommerce Returns?#

Ecommerce returns are online purchases a customer sends back to the retailer for a refund, an exchange, or store credit. The ecommerce return rate is the share of online sales, by units or by revenue, that comes back in a given period, and returns management is the process of receiving, inspecting, restocking, and refunding those items, and of learning why they came back.

TL;DR#

Online shoppers sent back an estimated 19.3% of what they bought in 2025, according to the National Retail Federation and Happy Returns 2025 Retail Returns Landscape report. Across all US retail, returns came to about $849.9 billion, or 15.8% of sales. The US Census Bureau put online sales at 17.1% of total retail sales in the second quarter of 2026, so the online return rate weighs more on retail results every year. Most retailers know their ecommerce return rate, but few know why each item came back, because the return reason dropdown records a label like "didn't fit" instead of a fixable cause. In the same NRF research, 82% of shoppers said free returns were a major factor in where they buy, which makes stricter policies a risky way to cut returns. The returns that stop are the ones whose cause gets found and fixed at the source: a size chart, a photo, a description, a box, or a carrier. A two to three minute conversation with each returner, run by an AI interviewer, surfaces that cause at the scale of every return.

This guide is for ecommerce, merchandising, and CX leaders who own the return rate and want it lower without making returns harder.

Ecommerce Returns Statistics for 2026#

Ecommerce returns statistics show a large and steady problem: close to one in five online purchases comes back. NRF and Happy Returns surveyed 2,006 consumers who had returned an online purchase and 358 ecommerce professionals at US merchants with more than $500 million in revenue. Their October 2025 report estimated:

Source: NRF and Happy Returns, 2025 Retail Returns Landscape.

The total rate dipped from 16.9% in 2024, according to the same research, but online returns stayed close to one in five. For a retailer selling $100 million online, that is roughly $19 million of merchandise coming back each year, before shipping, handling, and markdowns. Shoppers also expect returns to be easy: 76% preferred return options that offer an instant refund or exchange.

The takeaway for 2026 is that ecommerce is a growing share of retail and its return rate is structurally higher than the store's. Any retailer whose online mix is rising should expect returns to take a larger bite of margin unless the causes get fixed.

How to Calculate Ecommerce Return Rate#

Calculate your ecommerce return rate by dividing what came back in a period by what you sold in the same period, then multiplying by 100.

Return rate = (items or revenue returned ÷ items or revenue sold) × 100

Two choices make the number useful.

1. Pick units or revenue and stick with it. Unit return rate shows how often products fail customers. Revenue return rate shows the financial hit. Report both if you can, but never mix them in one trend line.
2. Break it down. Report return rate by product, by category, by size, by sales channel, and by carrier. A store-wide average of 18% can hide one product line at 40% and the rest at 10%.

Most retailers find that a small number of products drive a large share of returns. Those products are where the investigation should start, and they are the only place where a returns program will move the blended number quickly.

A useful companion number is the return rate for first-time buyers versus repeat buyers. A first order that ends in a return is a retention risk as well as a cost, which is why Ecommerce Customer Retention: Turning One-Time Buyers Into Repeat Customers treats the first return as a critical moment in the relationship.

Cost of Ecommerce Returns#

The cost of ecommerce returns is far larger than the refund, because every return carries a chain of costs and only some of them show up in the returns budget.

• Shipping both ways. The outbound delivery is lost, and the return label is often paid by the retailer.
• Processing. Someone receives, inspects, and sorts each item.
• Lost margin. Some items go back on sale at full price. Many are marked down, liquidated, or thrown away.
• Fraud. NRF and Happy Returns found 9% of returns were fraudulent in 2025.
• Lost repeat purchases. A customer who had a bad product experience may not buy again, even after a smooth refund.

Merchants in the NRF study named operational expenses (40%) and carrier shipping costs (40%) as their top returns cost drivers, followed by tariff uncertainty (33%). 64% said they planned to update their returns process within six months. Most of those updates focus on handling returns faster or cheaper. Fewer focus on preventing the next one.

That gap is the opportunity. A faster warehouse lowers the cost per return. Only a known cause lowers the number of returns. The last cost on the list, the lost repeat purchase, is also the one most often left off the returns budget, and it shows up later in ecommerce customer lifetime value rather than in the returns report.

Return Reasons the Dropdown Hides#

Return reasons collected through a dropdown hide the fixable cause, because the dropdown asks the shopper to sort themselves into a category the retailer wrote in advance. "Didn't fit" is a symptom. It does not tell the merchandising team which of these happened:

• The size chart ran small for that brand.
• The shopper ordered two sizes on purpose and kept one.
• The fit was fine, but the fabric felt different than the photo suggested.
• The item arrived late, after the event it was bought for.

Each of those has a different owner and a different fix. The size chart belongs to merchandising. Ordering multiple sizes is a policy and fit-confidence question. The fabric mismatch belongs to whoever writes product pages. The late delivery belongs to logistics. When all four become "didn't fit," nobody owns the problem and the return rate stays put.

Shoppers also pick whatever option gets the label printed fastest. In NRF's 2025 research, 45% of consumers said they believe "bending the truth" is acceptable when returning an item. A dropdown cannot follow up when a reason looks off. A conversation can, and it can do so politely, by asking what happened instead of challenging the answer.

This is the same problem forms create across the customer journey: people get flattened into a schema that was written before anyone knew what they would say. The principle underneath is that customer research in an AI-first business cannot start with a web form, and returns are one of the clearest places to see why. For a full bank of questions that get past the dropdown, see Post-Purchase Survey Questions That Actually Explain Returns. This post focuses on the program around those questions.

Returns Management Process That Reduces Returns#

A returns management process that reduces returns treats every return as product feedback as well as a logistics task. Fast handling keeps customers happy, but learning from each return is what lowers the rate next quarter. The five steps below sit on top of whatever returns platform you already run; if you are still choosing one, Best Returns Management Software in 2026 compares the options by how well they capture the reason behind each return.

Step 1: Find the Products That Drive Returns#

Find the products that drive returns by ranking every product by return rate and by total return cost. Start with the ten worst. Fixing one product at 40% returns does more than trimming the store average by a point, and it gives the team a visible win to build the program on.

Step 2: Ask Returners What Happened#

Ask a sample of returners for those products a few short questions in their own words. The Returns Experience Survey template does this as a conversation. The shopper picks a reason, and an AI interviewer follows up on it. If they say "didn't fit," it asks where it was tight or loose, what size they usually wear in that brand, and whether they checked the size chart. If they say "not as described," it asks which part of the description or which photo set the wrong expectation.

The conversation takes two or three minutes. Shoppers can answer by text or by voice in 57 languages, which matters for retailers selling across borders.

Step 3: Group Causes, Not Reason Codes#

Group returns by specific cause rather than by reason code, because only the cause points to a fix. Perspective AI turns each conversation into structured fields, such as product, stated reason, specific cause, and the team that owns it. Instead of "38% didn't fit," the merchandising team sees that most fit returns for one jacket came from shoppers who usually wear a medium and found the medium tight across the shoulders. That is a finding someone can act on this week.

Step 4: Route the Finding to the Owner#

Route each finding to the team that can fix it. Findings route to Slack, email, or your CRM. Size chart problems go to merchandising. Photo and copy problems go to ecommerce content. Damage patterns go to operations. Teams can also query the conversations directly from Claude and other AI assistants through the Perspective MCP server.

Step 5: Fix at the Source and Measure Again#

Fix the cause at the source, then measure the product's return rate again. Change the size chart, the photo, the description, or the packaging. Then compare the product's return rate for the 30 to 60 days after the change against the 30 to 60 days before it.

Shipping Damage and Delivery Problems#

Shipping damage and delivery problems cause a share of returns that the product team cannot fix, so they need to be separated from product causes early. An item that arrives damaged, late, or in the wrong box often comes back, and the dropdown records it as "damaged" or "no longer needed" with no detail about where the promise broke.

The Delivery Experience Survey asks customers what happened with their delivery and reports it by carrier and lane instead of one blended score. If one regional carrier crushes boxes or misses delivery windows, the pattern shows up in weeks instead of in the next quarterly carrier review. Pairing delivery feedback with return reasons separates product problems from shipping problems, which go to different teams. Given that merchants in the NRF study ranked carrier shipping costs (40%) among their top returns cost drivers, catching a bad lane early pays twice: fewer returns and fewer return labels.

Return Fraud and Bracketing#

Return fraud and bracketing are two different behaviors that both inflate the ecommerce return rate, and treating them the same way punishes honest shoppers. Return fraud means sending back something other than what was bought, or claiming a refund that is not owed; NRF and Happy Returns found 9% of returns were fraudulent in 2025. Bracketing means ordering several sizes or colors on purpose and returning the ones that do not work.

Bracketing is usually a fit-confidence problem, not a policy problem. A shopper who orders the same item in two sizes is telling you the size guidance did not give them enough confidence to pick one. Asking a bracketer why they ordered two sizes often surfaces the exact brand, cut, or fabric where the size chart falls short. Fix that, and the bracketing on that product drops.

Fraud needs a different response, but a conversation still helps. When a stated reason does not match the condition of the item, a follow-up question gives honest customers a chance to explain and gives the retailer a record. Combined with the 45% of shoppers who think bending the truth is acceptable, this is why a single dropdown answer is weak evidence for either fraud or fit.

How to Reduce Returns: A 60-Day Plan#

The way to reduce returns is to fix the causes behind your worst products first, one batch at a time. A 60-day plan keeps the work small enough to finish.

1. Weeks 1 and 2. Pull return rate and return cost by product for the last six months. Pick the ten products with the highest combined impact.
2. Weeks 3 and 4. Start a short returns conversation for those ten products. Aim for 30 to 50 conversations per product.
3. Weeks 5 and 6. Group causes by owner. Fix the three products with the clearest cause, whether that means a size chart, a photo, a description, or packaging.
4. Weeks 7 and 8. Move on to the next three products, and treat early numbers on the fixed ones as a first signal only, because most orders placed after the change have not reached the end of their return window yet. Compare like for like: the share of orders returned within the same number of days after delivery, before and after the fix. Read the final result once the full return window has closed on the post-change orders.

Retailers that also run stores have a second source of return reasons: store staff hear them at the counter every day. What store associates need to stay engaged covers how to capture what they hear. How this works across stores and online is on the retail and ecommerce industry page.

Returns Reduction Metrics to Track#

Returns reduction metrics show whether the program is lowering returns without hurting the customer relationship. Review these monthly.

• Return rate by product and category. The main outcome. Watch the worst products first.
• Return cost per order. Shipping, processing, and lost margin combined.
• Share of returns with a specific cause. The share of returns where you know the fixable reason beyond the dropdown label. This should rise as the program matures.
• Time from pattern to fix. How long it takes for a recurring cause to reach the owner and get corrected.
• Repeat purchase rate after a return. A guardrail. A returns process that frustrates customers lowers lifetime value even if the return rate drops.

The Repeat Purchase Research template asks customers what would bring them back after a first order, including one that ended in a return. For how returns feedback fits into the wider customer experience program, see Ecommerce Customer Experience in 2026: A Guide to Capturing the Why, and for the broader view of what shoppers want before they ever buy, see Shopper Insights in 2026: The Why Behind the Basket.

Frequently Asked Questions#

What is a good ecommerce return rate?#

A good ecommerce return rate is one that is falling for your own categories, because rates vary widely by product type. The National Retail Federation and Happy Returns estimated that 19.3% of US online sales came back in 2025, compared with 15.8% across all retail. Apparel and footwear usually run higher than that average, and consumables run far lower, so benchmark against your own history first.

How do you calculate return rate?#

Calculate return rate by dividing the value or number of items returned in a period by the value or number of items sold in the same period, then multiplying by 100. Track it by product, category, and sales channel, and keep units and revenue in separate trend lines. A single blended rate hides the few products that drive most returns.

What are the most common return reasons?#

The most common return reasons for online purchases are poor fit, the item not matching the description or photos, damage in transit, and changing one's mind. Those are the options shoppers pick from a dropdown, though. The cause behind them, such as a size chart that runs small or a photo that shows the wrong color, only comes out when someone asks a follow-up question.

How can retailers reduce ecommerce returns?#

Retailers reduce ecommerce returns by finding the specific cause behind the products with the highest return rates and fixing it at the source. That usually means correcting size guidance, rewriting product descriptions, replacing misleading photos, or improving packaging. Making returns harder tends to cost repeat purchases, since 82% of shoppers in NRF's 2025 research said free returns were a major factor when buying.

What is returns management?#

Returns management is the process a retailer uses to handle items customers send back, from the return request through receiving, inspection, restocking or disposal, and the refund or exchange. Good returns management also captures why the item came back and sends that finding to the team that can prevent the next return, such as merchandising, ecommerce content, or logistics.

What is return fraud?#

Return fraud is the act of returning merchandise to get a refund or credit that is not owed, such as sending back a different or used item. NRF and Happy Returns found that 9% of returns in 2025 were fraudulent. It is different from bracketing, where shoppers order several sizes on purpose, which is usually a fit-confidence problem rather than abuse.

Conclusion: Treat Ecommerce Returns as Product Feedback#

Ecommerce returns in 2026 run close to one in five online orders, and most retailers still manage them as a logistics cost. The return rate drops when each return is treated as product feedback: find the products that drive returns, ask the people returning them what actually happened, and send each specific cause to the team that owns it. Perspective AI does the asking at the scale of every return, with an AI interviewer that follows up on "didn't fit" until the cause is clear.

Start with your ten worst products. Start a Returns Experience Survey to get past the dropdown reason code, or see how Perspective AI supports CX teams turning return reasons into fixes.