Returns are no longer a minor operational cost for ecommerce.
They affect margin, warehouse capacity, customer service, fraud controls, inventory recovery and conversion. A generous returns policy can help a shopper buy. It can also create opportunities for abuse.
The strongest current benchmark comes from the National Retail Federation’s 2025 Retail Returns Landscape research with Happy Returns.
Ecommerce return and refund fraud statistics: key figures
| Metric | Latest benchmark | |—|—:| | Projected U.S. retail returns in 2025 | $849.9 billion | | Estimated online sales returned | 19.3% | | Estimated share of returns that are fraudulent | 9% | | Consumers saying free returns matter when shopping online | 82% | | Consumers saying some return-rule bending is acceptable | 45% |
These figures come from NRF’s October 2025 research.
How much merchandise gets returned?
NRF projected $849.9 billion in total retail returns for 2025.
That number includes physical and online retail. It demonstrates the scale of reverse logistics across the industry.
For ecommerce specifically, NRF estimated that 19.3% of online sales would be returned.
That makes online returns a core unit-economics problem, not an edge case.
What percentage of online sales are returned?
The NRF benchmark for 2025 was 19.3%.
The actual rate for one ecommerce company can be much lower or much higher.
Fashion, footwear and products with fit uncertainty can see higher return pressure. Consumables and simple repeat purchases can behave differently.
Do not apply 19.3% as a target to every store. Use it as market context.
How much return fraud is there?
NRF’s 2025 research estimated that 9% of all returns are fraudulent.
Return fraud can include several behaviors:
- returning stolen merchandise;
- returning a different item than the one purchased;
- using an item and returning it as unused;
- falsifying receipts or transaction information;
- abusing refund-without-return processes;
- manipulating delivery or return claims.
Retailers need to distinguish true fraud from ordinary customer mistakes.
Overaggressive fraud controls can damage legitimate customer relationships.
Why refund fraud matters more online
Ecommerce creates distance between purchase, customer and returned product.
A store may issue a refund before the warehouse fully inspects the item. A return can travel through a carrier or consolidated dropoff network. A marketplace seller may not physically see the returned product until days later.
That creates more points where data quality matters.
Order identifiers, serial numbers, package weight, product images, return reason, delivery scans and customer history can all help distinguish a normal return from suspicious activity.
Free returns still influence conversion
NRF found 82% of consumers said free returns are an important consideration when shopping online.
That creates tension.
Retailers want to reduce expensive and abusive returns, but making returns too difficult can reduce conversion before the purchase happens.
The best policy is rarely “make returns painful.”
The better goal is to remove unnecessary returns while using targeted controls for higher-risk behavior.
Return abuse is partly cultural
NRF found 45% of shoppers said it was acceptable to “bend the rules” when returning items.
That is a striking behavioral statistic.
Not every form of rule bending is organized fraud. Some shoppers may see wardrobing, late returns or vague return reasons as harmless.
Retailers still carry the cost.
This is why policy clarity matters. If customers do not understand the rules, enforcement becomes inconsistent and trust drops.
Return fraud vs return abuse
It is useful to separate the terms.
Return fraud involves deliberate deception for financial gain.
Return abuse can include exploiting a policy without necessarily using false identity or stolen merchandise.
Examples of abuse can include buying several versions purely to use free return shipping as a fitting room, repeatedly returning heavily used goods or exploiting lenient deadlines.
The operational response can differ.
Fraud may justify investigation or account restrictions. High-return behavior may be better addressed through product information, fit tools, better merchandising or differentiated return policies.
How returns affect ecommerce profit
A return can create costs far beyond refunding the sale.
Potential costs include:
| Cost | Why it matters | |—|—| | Outbound shipping | Already incurred | | Return shipping | Often retailer-funded | | Payment fees | Not always fully recoverable | | Warehouse labor | Receive, inspect, sort | | Refurbishment | Needed before resale | | Markdown loss | Returned item may no longer be full-price | | Customer service | Support time | | Fraud loss | Product or cash may be unrecoverable |
This is why a high-revenue store can still have weak economics.
For broader unit-economics context, see our ecommerce revenue and profit guide.
Why apparel returns are difficult
Apparel combines subjective fit, color expectations, size inconsistency and occasion-based purchasing.
A shopper can order several sizes because they cannot try the product in a store. That may be rational consumer behavior but expensive for the retailer.
Better size guides, customer photos, product measurements and fit recommendations can reduce avoidable returns without creating punitive policies.
Refund without return: useful but risky
Some retailers refund low-value items without asking for the product back because return shipping and processing would cost more than the item.
Economically, that can make sense.
It also creates an obvious abuse vector if customers learn that certain claims lead to automatic refunds.
The best systems use thresholds and risk signals rather than one blanket rule.
How marketplaces complicate return fraud
Marketplaces involve more parties.
The customer may buy from a third-party seller, use the marketplace payment system, return through a platform-managed workflow and receive support from the marketplace.
That can create disputes over:
- who approved the return;
- who received the returned item;
- whether the returned item matches the original;
- whether the seller met platform policy;
- who funds the refund.
Clear evidence matters more than informal communication.
What retailers can do to reduce return fraud
The strongest prevention systems combine policy, data and customer experience.
Identity and account signals can catch repeated abuse. Item-level identifiers help in electronics and high-value categories. Weight checks can flag empty-box returns. Warehouse photography can document mismatches.
But prevention also starts before checkout.
Accurate descriptions, strong product images, sizing information, compatibility tools and customer reviews reduce legitimate returns caused by bad expectations.
That lowers noise and makes truly suspicious behavior easier to identify.
Why blanket bans can backfire
A retailer can reduce returns by making them extremely difficult.
It may also reduce sales.
If 82% of consumers consider free returns important, aggressive restrictions can become a conversion problem.
A better model is segmented risk.
Low-risk loyal customers can get a smooth return experience. Repeated abusive patterns can trigger additional verification. High-value products can require serial-number checks.
That is more precise than treating every customer as a fraudster.
What should ecommerce teams track?
Return rate alone is not enough.
Track return rate by SKU, reason, customer cohort, channel and acquisition source. Add refund timing, recovered inventory value, return shipping cost and suspected fraud rate.
Then compare returned products with conversion performance.
A product with high conversion and high returns may have misleading merchandising. A product with low conversion and low returns may simply be unattractive.
Analytics should connect both sides.
Return fraud and customer trust
Fraud prevention is most effective when legitimate shoppers barely notice it.
A customer who has to prove innocence for every routine return may stop buying.
Retailers should make policies readable and explain exceptions clearly.
When a return is denied, the reason should connect to the policy and transaction evidence.
Returns as a merchandising signal
Returns data should not live only inside operations.
A high return rate can reveal a problem in product content, sizing, quality control, packaging or acquisition targeting.
Suppose one dress has a 34% return rate and the most common reason is “too small.” The right response may be to improve the size chart or product measurements rather than tighten the returns policy.
If one paid campaign produces customers with unusually high return rates, the ad may be attracting buyers with the wrong expectations.
That makes returns a feedback loop for merchandising and marketing.
The difference between gross return rate and net recovery
Two retailers can have the same return rate but very different economics.
Retailer A resells most returned items at full price.
Retailer B receives products late, damaged or out of season and has to liquidate them.
The headline return rate is identical. The financial result is not.
Useful operational metrics include:
- time from return request to resale;
- percentage returned to full-price inventory;
- percentage refurbished;
- percentage liquidated;
- percentage written off;
- cost per return;
- refund cycle time.
Those numbers convert a generic returns problem into an inventory-recovery problem.
Why faster returns can sometimes save money
A strict policy can look cheaper because it discourages returns.
But slow or confusing returns can keep products outside sellable inventory for longer.
In seasonal categories, speed matters. A winter coat returned in January can still be sold. The same coat returned in April may need a markdown.
That is why some retailers invest in consolidated dropoffs and faster reverse-logistics processing.
The goal is not simply fewer returns. It is higher-value recovery from the returns that still happen.
How return fraud affects policy design
Fraud controls should be strongest where the potential loss is highest.
A $9 accessory and a $1,500 laptop do not need the same return workflow.
High-value products may justify serial-number capture, stronger identity signals or inspection before refund. Low-value products may be cheaper to refund without reverse shipping.
Risk-based policy keeps friction proportional to exposure.
This is more sustainable than one harsh rule applied to every customer and every product.
The role of customer history
A retailer can learn more from patterns than from one return.
A customer who buys regularly and returns one damaged item is different from an account that repeatedly claims high-value packages were empty.
Historical signals can include return frequency, product mix, claim reasons, delivery disputes, account age and payment risk.
Those signals must be handled carefully so legitimate high-return customers are not automatically treated as fraudulent.
Fit-heavy categories can produce honest high-return behavior.
2026 takeaway for ecommerce teams
The 2025 NRF benchmark makes one point difficult to ignore: returns and return fraud are large enough to deserve executive attention.
The right operating model combines customer-friendly returns, better product information, fast inventory recovery and targeted fraud controls.
The companies that manage returns well will not necessarily have the lowest return rate. They will have the best balance between conversion, loyalty, recovery value and fraud loss.
Related ecommerce benchmarks
Return rate should be read alongside unit economics. Our ecommerce revenue and profit benchmark guide explains why strong topline sales can still produce weak owner economics when returns and fulfillment costs are high. For the security side of fraud and account risk, see our retail cybersecurity and breach statistics.
FAQ
What percentage of ecommerce sales are returned?
NRF estimated 19.3% of online sales would be returned in 2025.
What percentage of returns are fraudulent?
NRF’s 2025 Retail Returns Landscape estimated 9% of all returns were fraudulent.
How much merchandise is returned in the U.S.?
NRF projected $849.9 billion in total retail returns for 2025.
Do free returns matter to shoppers?
Yes. NRF found 82% of consumers said free returns were important when shopping online.
Is return abuse the same as return fraud?
Not always. Fraud involves deliberate deception, while abuse can include exploiting a policy without the same level of false representation.
Sources
- National Retail Federation, 2025 Retail Returns Landscape, October 15, 2025: https://nrf.com/research/2025-retail-returns-landscape
