Ecommerce Return Exception Workload Statistics 2026

Ecommerce return exception workload statistics need more than a return rate. A routine return may move through self-service authorization, carrier acceptance, receipt, and refund without an employee touching the case. An exception can require order research, policy interpretation, fraud checks, warehouse evidence, customer messages, and approval from someone with refund authority.

Public studies measure the scale of returns and some sources of pressure. They do not publish one reliable percentage for returns that need manual review. They also do not provide a cross-retailer average for review time or refund delay. This article keeps those gaps visible and shows how a retailer can calculate workload from its own queue.

Ecommerce return exception workload statistics at a glance

Sources: NRF and Happy Returns, October 2025 and Happy Returns, 2025.

The figures use different units. The 19.3% estimate is a share of online sales value, not a share of orders or units. The 9% figure is an industry estimate of fraudulent returns, not the percentage of cases sent to review. A merchant can review many legitimate returns and approve them.

Return volume is not exception volume

The National Retail Federation and Happy Returns projected that US retailers would receive $849.9 billion in merchandise returns during 2025, equal to 15.8% of annual retail sales. The estimate for online sales was 19.3%. The research included 358 ecommerce professionals at large US merchants with more than $500 million in revenue and 2,006 consumers who had made an online return during the previous 12 months.

These estimates describe returned sales value. A $500 item has five times the effect of a $100 item even if both require one case. Staffing plans therefore need transaction counts:

• return requests;
• returned units;
• parcels received;
• exceptions created; and
• manual touches recorded.

An exception rate needs a declared denominator. A retailer might calculate exceptions per return request, per returned unit, or per parcel. Each answers a different question. Keep the chosen definition stable across reporting periods.

The US Census Bureau reported quarterly ecommerce sales and ecommerce's share of retail sales, but its series does not report merchandise-return or exception rates. Census data can provide market context. It cannot supply the missing operational denominator.

No universal exception share exists

The reviewed NRF, Happy Returns, Appriss Retail, FedEx, Census, and academic sources do not publish a representative cross-retailer share of returns that require manual review. They also do not define an industry standard for what counts as an exception.

One merchant may classify a late return as an exception. Another may automate it under a customer-specific policy. Product category, item value, reason code, sales channel, carrier events, warehouse condition, payment method, and account history can all change the route.

A useful internal exception rate is:

return exception rate = return requests requiring manual action / all return requests

Report the rate by reason. A single blended number can hide whether the queue comes from damaged items, missing tracking, policy overrides, fraud screening, inspection failures, or refund errors.

Manual review creates several kinds of work

A manual review is not one task. The case may pass through support, warehouse operations, risk, payments, and finance. Each handoff adds queue time even when active handling takes only a few minutes.

Measure active handling time separately from elapsed resolution time. A reviewer might spend 12 minutes on a case that waits two days for warehouse photos. Both figures matter, but only one represents direct labor.

Fraud checks are only one part of the queue

The 2025 NRF and Happy Returns study estimated that 9% of all returns were fraudulent. Retailers that tracked specific patterns reported increases in overstated return quantities, empty packages, and decoy merchandise. The study did not say that 9% of every merchant's ecommerce returns were confirmed fraud cases.

Appriss Retail reported a broader 2024 estimate. Its analysis placed US merchandise returns at $685 billion and return and claims fraud and abuse at $103 billion, or 15.14% of returns. The analysis combined data from more than 60 large US retailers, Census data, a survey of 150 retail executives, and a survey of 1,000 consumers. It includes claims as well as merchandise returns, so it should not be compared directly with NRF's 9% estimate as if the difference were a trend.

A risk flag is not a confirmed case. Track the funnel in stages:

1. Returns screened.
2. Returns flagged.
3. Cases manually reviewed.
4. Cases approved, denied, or escalated.
5. Appeals and reversals.
6. Confirmed loss or recovery.

This prevents the screening rate from being reported as the fraud rate. It also shows whether tighter rules are increasing false positives, refund holds, and customer contacts.

Refund delays connect operations to customer contacts

Refund delay is usually elapsed time, not continuous staff effort. A return can wait for a carrier scan, warehouse receipt, inspection, approval, payment submission, or bank posting. The source studies reviewed here do not establish one universal average across those stages.

Consumer preferences still show why the delay matters. In the 2025 NRF study, 76% of consumers said they were more likely to choose a return option with an instant refund or exchange. The same study found that 71% would be less likely to shop with a retailer again after a poor return experience.

Use timestamps that identify the delayed step:

carrier wait = first carrier acceptance - return authorization

warehouse wait = warehouse receipt - first carrier acceptance

inspection wait = inspection decision - warehouse receipt

refund initiation wait = refund submitted - inspection decision

customer-visible refund time = refund confirmation - return authorization

Report the median and 90th percentile by return method and exception reason. A median can look healthy while a smaller set of old cases drives repeat contacts.

Contact workload needs its own denominator

A return does not always create a service conversation. One exception may create several. Customers can ask about eligibility, labels, tracking, receipt, inspection, refund timing, or an appeal.

The 2025 NRF consumer study found that 82% considered free returns important when shopping online. A separate FedEx and Morning Consult study, based on 2,200 US consumers surveyed in December 2024, found that two-thirds considered return policies before buying and 17% thought returns had become more difficult. These preference measures do not reveal a contact rate.

Retailers need two local measures:

exception contact rate = exceptions with at least one customer contact / all exceptions

contacts per contacted exception = return-related contacts / contacted exceptions

Tag contacts by stage and reason. If most contacts occur after warehouse receipt, the problem may be inspection or refund visibility rather than authorization policy. If contacts cluster before carrier acceptance, instructions or label delivery may be the source.

A transparent workload model

The following example is planning math, not an industry benchmark. Assume an online retailer handles 40,000 return requests in one month. Its own records show that 12% become exceptions, 30% of those exceptions receive a fraud check, and the remaining exceptions receive another form of manual review.

The 4,800 exceptions include 1,440 fraud checks and 3,360 other reviews. Manual review requires 840 hours:

(1,440 x 14 minutes + 3,360 x 9 minutes) / 60 = 840 hours

The contact assumptions produce 3,696 contacts and 431.2 handling hours:

4,800 x 55% x 1.4 x 7 minutes / 60 = 431.2 hours

Combined direct work is 1,271.2 hours. At 120 productive queue hours per full-time employee per month, the scenario requires about 10.6 full-time equivalents. A schedule would need to round up and account for hourly arrival patterns, leave, training, meetings, quality review, and peak coverage.

Replace every assumption with observed data. Do not use the 12% example as a published exception benchmark.

Seasonal pressure can move the bottleneck

The 2025 NRF retailer survey found that 43% planned to hire seasonal staff to handle returns and 49% planned to increase their focus on logistics partners. Happy Returns reported that 60% of retailers had faced a choice between shipping new orders and processing returns, while 68% considered returns-capability upgrades a priority for the next six months.

Those findings cover large US merchants and should not be applied as small-business averages. They do show that return labor competes with outbound fulfillment. When a warehouse prioritizes new orders, inspection queues can age. Customer service then receives more status and refund contacts even if its own process has not changed.

A staffing view should connect four queues:

1. Carrier exceptions waiting for investigation.
2. Received items waiting for inspection.
3. Risk cases waiting for a decision.
4. Approved refunds waiting for successful submission.

Count cases and age bands in each queue. A total backlog without stage data does not identify the staffing constraint.

Metrics for return exception staffing

Track volume, time, quality, and customer impact together:

Break the measures out by product category, return method, reason, risk band, sales channel, and decision. Record policy and system changes on the reporting timeline so a change in the rate has context.

Where support staff fit

Routine coordination can sit with trained support staff. Suitable work includes checking required fields, collecting photos, tracing carrier events, matching return authorizations to receipts, updating reason codes, sending approved status messages, and preparing complete case files.

Fraud determinations, policy overrides, high-value refund holds, account restrictions, and financial write-offs should stay with employees who have defined authority. Access should follow least-privilege rules, and the system should record each action.

Businesses that need coverage for the wider operation can review Stealth Agents' ecommerce staffing options and customer service support. The goal is to give routine cases a clear owner while preserving an authorized escalation path.

Frequently asked questions

What percentage of ecommerce returns become exceptions?

No representative public source reviewed for this article gives a universal percentage. Calculate the rate from return requests that required manual action, and publish the exact rule used to classify an exception.

How many online sales are returned?

NRF and Happy Returns estimated that 19.3% of US online sales would be returned in 2025. The estimate measures sales value, not the share of orders, items, or parcels.

What percentage of returns need a fraud check?

The public sources do not provide a universal manual fraud-review share. NRF estimated that 9% of all retail returns were fraudulent in 2025, but that estimate is not a screening rate. A merchant's risk rules may flag a larger or smaller share, including legitimate returns.

How should a retailer measure refund delay?

Record authorization, carrier acceptance, warehouse receipt, inspection decision, refund submission, and customer confirmation timestamps. Report median and 90th-percentile time by stage and exception reason.

How can a retailer forecast return exception staffing?

Multiply return requests by the observed exception share and average handling minutes. Add customer contact volume and handling time. Divide total hours by productive queue hours per employee, then add coverage for peaks, leave, training, and quality work.

Can customer service staff manage return exceptions?

They can handle documented administrative steps such as evidence collection, status updates, carrier tracing, and case preparation. Decisions involving fraud, policy overrides, account action, or high-value refunds need an authorized owner.

Sources and limitations

• National Retail Federation, Consumers Expected to Return Nearly $850 Billion in Merchandise in 2025, October 15, 2025. Joint NRF and Happy Returns research with 2,006 consumers and 358 ecommerce professionals at large US merchants.
• National Retail Federation, 2025 Retail Returns Landscape. Research landing page for the return estimates, consumer findings, fraud estimate, and retailer responses.
• Happy Returns, 2025 Retail Returns Landscape. Companion report page used for return-method preferences and retailer operating-pressure findings.
• Appriss Retail, Fraudulent Returns and Claims Cost Retailers $103 Billion in 2024. Analysis using data from more than 60 large retailers, Census data, 150 retail executives, and 1,000 consumers. Its scope includes claims and abuse.
• FedEx, Second Annual Returns Survey, January 28, 2025. Morning Consult survey of 2,200 US consumers and 1,000 US business shippers conducted in December 2024.
• US Census Bureau, Quarterly Retail E-Commerce Sales. Official US sales series that does not measure merchandise returns or exceptions.
• Hjort and colleagues, Fraudulent Returns in Omnichannel Retailing. Peer-reviewed qualitative research on return-fraud practices and measurement limits.

The 2026 title identifies the review edition. The article does not present the planning example as observed industry performance. Public research supports the scale and pressure findings, while each retailer must measure its own exception share, manual-review time, refund delay, and contact workload.