Nine and a half days. That is the median gap between a delivery and a return for an ordinary parcel, and it nearly doubles to eighteen days for returns that later look fraudulent. As of November 14, 2026, that gap is the most useful number in peak planning, because the marketplace refunds on one clock while the parcel travels on another, and the difference between the two is where margin quietly leaves the building.
Two clocks, one parcel
The refund clock is a service promise and the parcel clock is a physical process, and platforms have spent the past two years shortening the first while the second stayed roughly the same. A membership tier that refunds before the parcel is dropped off makes the gap wider, not narrower, because the money moves at the moment of the claim rather than at the moment of receipt. The retail dispatch for November 13 notes exactly this pattern in the new paid tier being tested by TikTok Shop, where eligible members can receive store credit before handing the item back.
That is a reasonable loyalty feature and a difficult operational one. It converts returns speed into a benefit, which raises the cost of getting the return wrong, and it pushes the verification step later in the process. For the seller, the practical consequence is that the cost of a bad return is no longer bounded by the value of the goods, because the refund has already been issued by the time the parcel is inspected.
Fraud looks small until it is scaled
Five percent sounds manageable in a slide. The IMRG round-up runs the arithmetic and it stops sounding manageable: at a 100 million pound turnover, a 5 percent fraud rate is about 1 million pounds a year; for a large omnichannel retailer it rises to 3.5 million, and at enterprise scale to 20 million. The multiplier is not the fraud rate, it is the sales base the rate is applied to.
| Retailer scale | Annual sales | Fraud rate | Estimated annual loss |
|---|---|---|---|
| Mid-market | 100 million pounds | 5 percent | About 1 million pounds |
| Large omnichannel | Several hundred million | 5 percent | About 3.5 million pounds |
| Enterprise | Multi-billion | 5 percent | About 20 million pounds |
The same analysis offers one behavioural signal that costs nothing to instrument: the longer a customer holds a product before returning it, the higher the risk. That is why the lead time gap matters. A rule that flags returns older than a fortnight will catch some genuine late returns as well, so the threshold has to be tuned against the false positive rate rather than against the fraud rate alone.
The peak calendar moved, the return window did not
Promotions started earlier this year, which spreads the order curve but does not move the return window. Walmart's first Black Friday event ran from November 12 to 15, Amazon's Black Friday Week begins November 19, Best Buy started on November 6 and Target ran an early event from November 5 to 7, according to the same dispatch. The result is a longer selling period followed by a return wave that still lands in the same few weeks of January.
Why an earlier peak is harder on returns
An early purchase sits in the customer's home for longer before the return decision is made, which stretches the average holding period and makes the lead-time signal noisier. It also means the item has been out of the fulfilment centre for longer, so the chance that it comes back in resaleable condition drops, and the recovery value falls with it.
Cost to serve has to be counted per order
Peak planning usually budgets revenue and headcount. The number that decides whether the season was profitable is cost to serve per order, which includes the outbound leg, the pick and pack, the inbound leg if it comes back, the inspection and the markdown if it cannot be resold. Averages hide this, because a 20 percent return rate applied to a mixed basket says nothing about which orders are losing money.
Duty stacking changes what a return is worth
Cross-border returns now carry a second layer of cost. The European Union ended the low-value consignment exemption in July 2026 and added a handling fee in November, and the first full quarter of data showed Shein, Temu and AliExpress combined German sales down about 35.5 percent to 566 million euros, with their share falling from 5.3 to 3.1 percent, according to B2B Daily. A returned low-value item that has already paid duty and a handling fee is worth less than the same item returned inside a single market.
That changes the return policy arithmetic for cross-border sellers in a way that is easy to miss. If the item is below the threshold where resale recovers the duty, the rational move is to write it off rather than ship it back, and the policy has to say so explicitly. Without that rule, the reverse leg gets paid twice: once in freight and once in duty that cannot be reclaimed.
What to instrument before December
The measurement list is short and most of it can be built from data the operation already holds. None of it requires a new platform, and two of the five items are usually sitting unused in the warehouse log. What matters is that the five measures are read together, because each one alone can be gamed by a channel decision that simply shifts the cost somewhere else. Read as a set, they show whether a refund is being paid once or twice.
The second item on that list is the one most often missing. Without a recorded lead time, a returns team can see that fraud is happening but cannot separate a slow genuine customer from a fast fraudulent one, and any policy it writes will be a guess dressed up as a rule. Review analysis and consumer feedback help here as well, because the stated reason on a return often differs from the pattern in the timestamps.
FAQ
Is instant refund membership actually a returns problem?
It shifts the risk rather than creating it. The refund is issued before inspection, so the seller carries the exposure for longer and has less leverage if the item comes back unsaleable.
Can lead time alone identify fraudulent returns?
No, but it is a cheap signal. It needs a threshold tuned against false positives, and it works better combined with category and value than on its own.
Why does an earlier peak make returns harder?
Items sit with customers for longer, so the holding period stretches, the lead-time signal gets noisier and the chance of resale at full value falls.
Should cross-border returns always be shipped back?
Only when resale recovers the duty already paid. Below that point, writing the item off is usually cheaper than paying the reverse leg twice.
