Returns Season: Support Staffing for a 15.8% Return Rate
- Retail, BPO
- 15 Views
Retail returns season A support event with a predictable date, a predictable volume and a predictable shape. The retail returns season lands in the month support teams are smallest. $849.9bnReturned to US retailers in 2025, 15.8% of annual sales (NRF) 19.3%Of online sales came back, against 15.8% across all channels 9%Of all returns were fraudulent, […]
Retail returns season
A support event with a predictable date, a predictable volume and a predictable shape. The retail returns season lands in the month support teams are smallest.
Returns get planned by the logistics team and absorbed by the support team. Warehouse capacity, carrier contracts, restocking labour and disposition rules all get a budget line and a project owner. The contact volume that every return generates gets handled by whoever is left on the rota in January.
The numbers are not small. The National Retail Federation’s 2025 Retail Returns Landscape, produced with Happy Returns, put total US returns at $849.9 billion, equal to 15.8% of annual sales. Online returns ran at 19.3%. Retailers expected 17% of holiday sales specifically to come back. Those figures are for 2025; the 2026 edition had not been published at the time of writing.
This post treats returns as what they also are: a support event, with a predictable shape, landing in the month most retail support teams have just shrunk.
What a single return actually generates
A return is rarely one contact. It is a chain, and the length of the chain is a design decision rather than a fact of nature.
Six stages, each capable of generating a contact. A well-designed returns experience produces perhaps one contact per ten returns. A poorly designed one produces two or three contacts per return. The difference between those two operations is not staffing, and no amount of support headcount closes it.
Why January is the worst-staffed month in retail support
The sequence is familiar and it repeats every year.
- Seasonal agents are recruited in September and October and trained through November.
- Peak sales volume lands from late November through December, and the team is sized for it.
- Contracts end in the first or second week of January, because the sales peak has passed.
- The returns wave arrives from late December through February, into a team that has just lost a third of its capacity.
The mismatch exists because returns coverage is planned against the sales calendar rather than the returns calendar. Returns lag purchase by 20 to 60 days depending on your return window, which puts the peak of the returns contact wave several weeks after the peak of the sales event that caused it.
The planning fix is one line in the contract
If seasonal capacity is staffed at all, staff a portion of it to the end of February rather than the second week of January. It costs less than rehiring, and it keeps trained agents on the contacts that are hardest to handle cold: refund timing and condition disputes involving a customer who has already waited.
Return fraud, and the agent caught in the middle
NRF’s 2025 research found 9% of all returns were fraudulent. Appriss Retail separately put the cost of fraudulent returns and claims at $103 billion in 2024. The same NRF research found 45% of consumers think it is acceptable to bend the rules on returns, and 82% say free returns matter to where they shop.
Those figures describe a genuine tension rather than a solvable problem. Tighten the policy and you lose customers who value free returns. Loosen it and you fund the 9%. The support team lives inside that tension on every call.
What to give agents instead of discretion
- A decision rule, not a judgement call. Define the value and frequency thresholds at which a return goes to review, and let agents apply the rule rather than assess the customer.
- Customer history on the screen. An agent deciding a condition dispute without return history is guessing. This is an integration requirement, not a training one.
- A named escalation route that resolves inside 24 hours. Reviews that sit for a week convert a suspicious return into a furious customer who may well have been legitimate.
- Explicit permission to approve below a threshold. The cost of investigating a low-value return usually exceeds the value of the item. Say so in policy rather than leaving agents to work it out.
What returns costs to process, and why the support line is invisible
The most cited figure on returns processing cost comes from a CBRE and Optoro study reported in December 2021: $33 to process a return on a $50 item, or 66% of the item’s retail price. It is five years old and should be labelled as such whenever it is used. Newer per-vertical tables circulating online have no research behind them and should not be cited at all.
What no published figure covers is the support cost inside that number. Two to four contacts per return, at a fully loaded cost per contact of around $7, is $14 to $28 of support cost on a single return. On a $50 item, that is potentially more than the logistics cost of moving it back. Our piece on cost per contact shows how to calculate your own figure, so you can put a real number against your own return volume.
Automate the queue, keep the exceptions
| Contact type | Automate? | Reasoning |
|---|---|---|
| Return eligibility and window check | Yes | Deterministic. The answer is in the order record |
| Return initiation and label generation | Yes | High volume, no judgement required, and customers prefer self-service for it |
| Return receipt confirmation | Yes, proactively | Removes the contact rather than handling it |
| Refund status | Yes for status, no for disputes | Status is a lookup. A dispute about timing is a retention conversation |
| Condition disputes | No | Judgement, evidence, and a customer who feels accused |
| Fraud review outcomes | No | Never deliver a fraud decision through automation |
| Second contact on any of the above | No | A repeat contact means the first attempt failed. Route to a tenured agent |
Metrics that tell you whether returns support is working
| Metric | Definition | What it exposes |
|---|---|---|
| Return contact ratio | Support contacts divided by returns processed | Whether your returns experience is self-service or not. Target well below 1.0 |
| Refund cycle time | Return receipt to refund issued | The single largest driver of returns contact volume |
| Repeat contact rate on returns | Share of return contacts that are a second or later touch | Whether the first contact actually resolved anything |
| Return contact cost per order | Return contact volume x cost per contact, divided by orders | Lets you put returns support into unit economics alongside logistics |
| Dispute rate and overturn rate | Share of returns disputed, and share of disputes decided for the customer | A high overturn rate means your decision rules are wrong, not that agents are soft |
A retail returns season coverage plan, in six decisions
| January returns readiness | Rate 1 to 5 |
|---|---|
| Returns contact volume forecast built from the sales forecast plus your actual return rate and return window | 1 2 3 4 5 |
| Seasonal capacity end date set against the returns wave, not the sales peak | 1 2 3 4 5 |
| Refund cycle time measured, with a target, and owned by a named person | 1 2 3 4 5 |
| Self-service returns initiation live and tested at peak load | 1 2 3 4 5 |
| Proactive return receipt notification live | 1 2 3 4 5 |
| Fraud decision rules documented, with thresholds, so agents apply rather than judge | 1 2 3 4 5 |
| Return history visible on the agent desktop | 1 2 3 4 5 |
| Escalation route for disputes resolving inside 24 hours | 1 2 3 4 5 |
| Repeat-contact routing to tenured agents configured | 1 2 3 4 5 |
| Return contact ratio and refund cycle time on the January reporting pack | 1 2 3 4 5 |
Retail returns season: the short version
Returns are a support event with a predictable date, a predictable volume and a predictable shape, and they land in the month retail support teams are smallest. Forecast the wave from your own return rate and return window, keep trained capacity through February rather than January, remove the four contact stages that are removable by design, and keep the two that are not in front of your most experienced agents.
Returns contact volume estimator
Returns are planned as a logistics cost and absorbed as a support cost. This estimates the support side: how many contacts your returns generate, what they cost, and how many agents the January wave needs.
| Measure | Annual | January wave |
|---|
Plan your January returns coverage
Plan your January returns coverage with Centro. Bookings close 15 December for a January start.
Sources
- National Retail Federation and Happy Returns, 2025 Retail Returns Landscape, published 15 October 2025. https://nrf.com/research/2025-retail-returns-landscape
- NRF press release, Consumers Expected to Return Nearly $850 Billion in Merchandise in 2025. https://nrf.com/media-center/press-releases/consumers-expected-to-return-nearly-850-billion-in-merchandise-in-2025
- Appriss Retail, fraudulent returns and claims cost, $103 billion in 2024. https://www.businesswire.com/news/home/20241230601195/en/
- CBRE and Optoro, returns processing cost, reported December 2021. Five years old, label the date when citing. https://chainstoreage.com/holiday-returns-soar-and-so-are-retailers-return-costs
- US Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026, and Occupational Outlook Handbook. Basis for the cost per contact figure. https://www.bls.gov/news.release/ecec.t04.htm