Most writing about returns stops at the customer's screen: the portal, the label, the policy. The expensive part happens afterwards, on a bench in the warehouse, where a parcel with no order number on the outside has to be opened, matched, judged and sent somewhere. This page is about that part.
Short answer: returns processing is the sequence from a returned parcel arriving at the dock to the item reaching its final destination and the customer receiving their outcome. The steps are receiving, matching the parcel to a return request, inspection and grading, disposition (restock, refurbish, resell as second quality, donate, recycle, return to vendor), the refund or exchange trigger, and the record that ties it all back to the original order. The two decisions that cost money are when to release the refund and where the item goes. The rest is throughput.
The seven steps
| Step | What happens | Where it goes wrong |
|---|---|---|
| 1. Receiving | Parcel is scanned in at the dock against the return label's tracking number | Parcels with no pre-registered return, or a customer's own label, sit in a pile |
| 2. Matching | The tracking number resolves to a return request and its original order | The return exists in the returns tool but not in the warehouse system, so the bench works from a printed slip |
| 3. Inspection | Item is checked against the reason the customer gave and the condition rules | Inspection standards differ by person; "unworn" means different things on Monday and Friday |
| 4. Grading | Item is assigned a grade: A (as new), B (minor wear), C (damaged), faulty | No grades at all, so every return is either restocked or binned |
| 5. Disposition | Grade plus product rules decide the route: restock, refurbish, second-quality resale, donate, recycle, return to vendor | Decided ad hoc, so resale value is lost and restock quality slips |
| 6. Outcome trigger | Refund, exchange dispatch or store credit is released | Released on receipt for everyone, or held until inspection for everyone, instead of by rule |
| 7. Record | The return, its grade, disposition and cost are written back to the order | Nothing is written back, so return-reason analytics are guesswork |
The decisions that matter
When to release the refund. Three policies exist: on the customer's request (instant), on the carrier's first scan of the return label (in transit), or after inspection (on receipt). Instant is best for the customer and riskiest for the brand; on receipt is safest and slowest. The useful answer is a rule per segment: instant for customers with a clean return history below a value threshold, on scan for most, on inspection for high-value items and for accounts flagged by return fraud signals. A rule is only possible if the returns tool and the warehouse share the order record, because "after inspection" has to mean the inspection result reached the system that pays.
Where the item goes. Disposition is a margin decision. An A-grade item back on the shelf within two days has nearly full value; the same item three weeks later at the end of a season has a fraction. Rules by category and grade ("A-grade apparel restocks; B-grade goes to the outlet channel; C-grade and faulty are photographed and recycled or returned to vendor") make the decision fast and consistent. Some items should never come back at all: the case for that is in returnless refunds.
Exchanges first. If the customer asked for an exchange, the replacement should ship on the request or on the return scan, not after inspection, or the customer waits two weeks for a size swap. This is where exchange-first policies earn their keep; the commercial argument is in how to increase exchange rate.
What to automate and what to keep human
Automate: receiving scans, matching to the request, the refund trigger by rule, the disposition route by grade and category, the write-back to the order, the customer message at each step ("we have received your return", "your refund is on its way"), and the exception list for parcels that arrived without a request.
Keep human: the grading judgement itself, disputes where the customer's stated reason and the inspection disagree, and the periodic review of grading standards so that three inspectors grade the same item the same way.
Instrument: time from dock scan to disposition, time from disposition to refund, share of returns by grade, share restocked versus resold versus written off, and cost per return including the inbound label. The true cost of a return shows how those costs add up, and return reason analytics shows what to do with the reasons.
The warehouse and the returns tool must share one record
The common failure is structural, not procedural. The returns portal creates a request; the carrier moves the parcel; the warehouse receives it on a different system; finance refunds on a third. Each step works. The handovers do not. The bench cannot see the request, finance cannot see the grade, the customer cannot see any of it.
Pango's design puts the return on the same order record as the outbound shipment and the warehouse pick and pack, so a returned parcel is recognised at the dock by the label Pango generated, the inspection result is written to the order, and the refund or exchange rule fires on that result. Rules are written in plain language and compiled into a per-merchant workflow, including the disposition routes and per-country refund logic; complex cases (partial refunds, buy-X-get-Y unwinds, country-specific rules) are built to fit the merchant's policy rather than configured from a fixed menu. The module is described at return management, and the wider operating model in autonomous returns with AI agents.
A worked example
A customer requests a return of two items, one for size and one for damage, and asks for an exchange on the size. The request creates a return with two lines and a replacement order for the new size, held. The customer prints the label. The carrier's first scan releases the replacement (exchange-first rule) and the customer is told it is on its way. At the dock, the parcel is scanned and matched. The size item grades A and restocks the same day; the damaged item grades C, is photographed, and routes to return-to-vendor with the photo attached. The refund for the damaged item fires on the C grade, per the rule "faulty or damaged refunds on inspection, with the inbound shipping cost refunded too". Both grades, both dispositions and the cost of the inbound label are written to the original order. The reason codes feed the monthly review that shows this product's damage rate is three times the category average, which is a packaging problem, not a returns problem.
The bottom line
Returns processing is where a return becomes either a restocked item and a retained customer or a written-off item and a refund. Define grades, write disposition and refund rules per segment, automate the handovers, keep the judgement human, and make sure the bench, the portal and finance are looking at the same order. If your returns tool and your warehouse do not share a record today, that is the first thing to fix, and the shipping and returns audit is where to start. Or book a demo to see the workflow above run on one record.
