Ecommerce returns solutions, and the half they can't fix

What an ecommerce returns solution actually is

An ecommerce returns solution is the set of systems a brand uses to receive, decide on and settle a returned order: the customer-facing request and label, the reverse transit, the inspection, and the refund or exchange that closes it out. Most of what is sold under that name solves the paperwork of a return rather than the return. Some of the returns hitting a portal are the customer's own choice, and a portal helps with those. The rest are broken promises wearing a returns label, and a portal removes none of them, because it only starts working after the promise is already broken. I closed Keeyu's first thirty-five customers myself, and the same shape came up in nearly every one: returns volume climbs, the returns desk gets blamed, and the cause sits two steps upstream in the pick or the delivery run.

The volume isn't in dispute. NRF and Happy Returns put returned merchandise at $849.9 billion in 2025, with 19.3% of online sales sent back. The Census Bureau put US e-commerce at $326.7 billion in the first quarter of 2026. That's why post-purchase operations now gets its own budget line.

Split your return rate before you buy anything

Your return rate isn't one number. It's two numbers sharing a form field, and they answer to completely different spending.

Demand side is wrong size, changed my mind, didn't suit, bought three to keep one. The customer chose this. Better sizing data and a clean self-service portal move the number, and this is the side the market writes about.

Supply side is wrong item picked, damaged in transit, duplicate shipment, or a delivery exception that held the parcel long enough for the customer to buy it elsewhere. Nobody chose this. A portal can't reduce it by a single unit, because it opens only after the promise is broken, which is the same reason a WISMO ticket never made a late parcel arrive.

A third bucket belongs to neither half: the same NRF and Happy Returns research puts 9% of returns down as fraudulent, with 45% of shoppers saying that bending the rules is acceptable.

So pull last quarter's returns and tag each one demand or supply, then buy against the split rather than the total. If eight in ten are demand side, a portal is a good purchase. If the split runs the other way, a portal makes the paperwork faster and leaves the volume where it is.

The four kinds of returns solution, and what each one cannot fix

Four categories cover almost everything sold as a returns solution. Each does something real, and each has a sentence the vendors leave off.

Platform-native returns. Your commerce platform already models this: the documentation for building return management treats the reverse fulfillment order as the unit of work. It covers the standard path at no extra cost and stops at the first exception.

Returns portals and returns apps own the customer-facing request, the label, the exchange offer and the store credit. They genuinely move demand-side returns and self-service rates, and do nothing to supply-side volume. The case studies quote a blended rate for a reason.

Reverse logistics and 3PL services take the physical side: transit, receiving, inspection, restock and disposition, plus the carrier label services USPS sells to merchants. Priced per unit handled, so it makes each return cheaper and never reduces how many there are.

Proactive e-commerce operations watches the order after checkout, detects the break, decides what to do and acts. It reduces the supply-side share by preventing the return, and it's the only one of the four that can.

We're the fourth item, and I'll be exact about what that excludes. Keeyu isn't a returns portal, not a returns platform, not a carrier, not a 3PL and not an order management system. If you need a branded page where a customer picks a reason code and prints a label, buy one. We sit upstream of that. One in five orders hits an operational break after checkout, and our job is to catch it before it becomes a return.

The returns process, and the three places it breaks

Whatever tooling sits on top, the process underneath runs four steps.

1. Request and approval.
2. Label and reverse transit.
3. Receipt and inspection.
4. Disposition and settlement.

Three of those steps are where the cost actually accumulates.

Approval sits in a queue. Platform documentation is explicit that creating a return opens it in a state that assumes approval already happened somewhere else, which in most brands means a person and a spreadsheet. Transit is dark. A standard ground return spends two to five days in the network, and the customer refreshes an inbox for every one of them.

Inspection is where supply-side returns finally become visible, days after the break that caused them, to the one team with no power to fix it. An RMA in shipping record tells you a unit came back. It doesn't tell you the pick was wrong.

Detect, decide, act is the shape of the fix, and the reason the returns process feels so expensive is that all three of those currently happen at step three. Detection should have happened before step one.

The decision nobody automates: what to do with this return

Once the unit is coming back, somebody decides what happens to it. The option list is public. Making the right choice per order is what nobody has automated.

• Refund on receipt: the default. Safe, slow, and the customer waits out transit plus inspection.
• Refund on scan: settle when the carrier scans the parcel, not when the warehouse opens the box. Trades a small loss rate for days of silence.
• Returnless refund: refund and let the customer keep the item when recovery value sits below handling cost. Textiles are the standing case: the EPA counts 11.3 million tons landfilled in its reference year, and shipping a garment back to be destroyed spends money to make waste.
• Exchange or store credit: keeps the revenue inside the business instead of reversing it.
• Hold for inspection: for high-value units or suspected fraud only. It should be the exception the system escalates, not the default it falls back to.

There's a deadline on all of it. Once the buyer's right to a refund vests, the FTC's Mail, Internet, or Telephone Order Merchandise Rule requires a prompt refund: within seven working days where the customer didn't pay on credit you extended, and within one billing cycle where they did. 16 CFR 435 is the text that says so. The decision itself is the product, and it should be made against the order's own fulfillment record rather than a static policy rule: a unit that came back because the warehouse picked the wrong size doesn't get the same decision as one the customer ordered in three sizes.

What to measure so the solution has to prove itself

Four metrics, picked so no vendor can take credit for a change it didn't cause.

1. Return rate split by cause: never the blended rate, which is what lets a portal take credit for a change in the mix.
2. Time from carrier scan to refund sent: measured against the FTC's refund deadline for how your customers actually pay, not a service level you set yourself.
3. Revenue retained as exchange or credit: as a share of returned value.
4. Cost per return processed: given as components rather than a headline number: reverse transit, plus handling labor, plus disposition loss. Anchor the labor component in the BLS median annual wage for hand laborers and material movers, $37,680 as of May 2024. That's a component and not a total, and every per-return dollar figure circulating on this topic traces back to a vendor.

Then run one test at year end. If your supply-side rate hasn't moved after twelve months on a returns platform, that's not a failure of the platform. It was never the thing that was broken.

Stop paying for returns you caused

You now know which part of your return rate a portal can't touch. The supply-side share was created upstream, in a pick, a pack or a delivery run, long before anyone opened a return request. Keeyu is the operations layer for that share: we watch every order after checkout, detect the break, decide what to do and act, usually before the customer knows anything went wrong. If your supply-side returns aren't falling, book a demo and bring last quarter's split.

• National Retail Federation and Happy Returns. 2025 Retail Returns Landscape. Returns of $849.9 billion in 2025, 19.3% of online sales returned, 9% of returns fraudulent and 45% of shoppers finding rule-bending acceptable.
• National Retail Federation. Returns projected at $849.9 billion in 2025. The all-channel return rate of 15.8%, against the higher online-only figure.
• US Census Bureau. Quarterly Retail E-Commerce Sales, Q1 2026. E-commerce sales of $326.7 billion, 16.9% of total US retail.
• Shopify. Build for return management. Official platform documentation for what native returns handling covers.
• Shopify. Manage reverse fulfillment orders. The reverse fulfillment order as the unit of returns work on a commerce platform.
• Shopify. returnCreate mutation. A return is created in an open state, with approval assumed to have happened elsewhere.
• United States Postal Service. Customer returns and label services. The carrier-side return label options a merchant buys per unit.
• United States Postal Service. USPS Ground Advantage. Two to five day transit, which is what a returning parcel spends in the network.
• Federal Trade Commission. Mail, Internet, or Telephone Order Merchandise Rule. The obligation to refund promptly: seven working days where the buyer did not pay on credit the seller extended, one billing cycle where they did.
• Electronic Code of Federal Regulations. 16 CFR 435.2. The substantive prompt-refund rule, with the deadlines defined at 435.1(b).
• US Environmental Protection Agency. Textiles: Material-Specific Data. 11.3 million tons of textiles landfilled in the reference year.
• US Bureau of Labor Statistics. Hand Laborers and Material Movers. Median annual wage of $37,680 as of May 2024, the labor component of processing a return.