Returns Evidence Guide

Empty-Box and Wrong-Item Returns: An Evidence Checklist for Brands and 3PLs

By EasyShippingX Editorial TeamPublished Updated

A return lands on the receiving bench. The tracking page says Delivered. The label matches an approved RMA. The customer service queue already holds a message asking when the refund will go out. Then the box is opened and it is empty, or it holds two units instead of four, or it holds something the customer never bought from you.

What follows is rarely a decision. It is a three-way stalemate. The warehouse has a discrepancy note and a box nobody is allowed to move. Customer service has a customer who is now angry about a second delay. Between them, the brand has to decide whether this goes to the customer, the carrier, or the 3PL, and nobody in the chain is holding evidence strong enough to force that decision.

This guide starts at that moment, the first hour after an anomaly is found, and works forward. It does not start from a definition of return fraud, because most anomalous returns that cost a brand money are not fraud. They are mislabeled parcels, carrier mishandling, a customer who grabbed the wrong box off the counter, a pick error nobody caught, and a genuinely empty carton that was never full. Starting with fraud makes every case look adversarial, and adversarial cases are the ones that end in a refund paid twice.

The order of work matters more than the enthusiasm behind it. You establish handling discipline and measurement conventions first, then let the evidence decide which path a case belongs to.

Editorial illustration of a warehouse return station at the moment of unpacking, an open cardboard box with crumpled packing paper beside a small bench scale and a clipboard, one worker pausing mid-motion with a phone raised to photograph the contents, warm overhead light, generous empty space at the top of the frame, visible paper grain and slightly imperfect ink linework.
The first hour: do not move the box yet

Why tracking, weight, and photos prove different things

The instinct when a dispute opens is to grab the tracking record, because it is the one piece of documentation both sides can see. It is also the weakest piece. A Delivered scan proves that a parcel reached an address. It says nothing about what was inside, whether the seal was intact, or whether the box that arrived is the box that shipped.

Weight is more useful, but only if you keep three different numbers apart. The declared weight printed on the label is what the shipper entered, often rounded and sometimes estimated. The billed weight is what the carrier charged for, which may include dimensional weight and is a pricing construct rather than a measurement of the parcel. The measured weight is what your scale reads on the receiving bench, at a known time, on a known device. Mixing these produces arguments that cannot be resolved, because a billed weight is not evidence about contents and a declared weight is not a measurement.

Two more limits are worth stating plainly. Matching weights does not prove matching contents; a box of the right mass can hold the wrong thing, so weight is a filter for theories, not a verdict. And a contents mismatch on its own does not assign fault, because an empty box is equally consistent with a customer error, a carrier loss, and a packing station that closed a carton before the item went in.

The practical conclusion is a separation of two activities that teams habitually merge: finding the anomaly and attributing the cause. The first is a mechanical observation anyone can record in minutes. The second requires a specific piece of independent evidence, and if that evidence does not exist yet, the honest output of the investigation is a case marked undetermined with a follow-up date, not a guess with a signature on it.

The intake sequence: six steps from anomaly to owner

The six steps below are deliberately ordered, because each one protects a piece of evidence the next one depends on. Skipping step two, for example, by moving the parcel into normal stock, destroys the ability to do step five at all.

First, confirm parcel identity. Match the physical parcel to the original order, the customer or tenant, the RMA, the tracking number, and the receiving record. This step exists because an alarming share of disputes turn out to be a mapping error: the box is fine, the return authorization belongs to a different order, and the customer is owed a routine refund with no investigation at all.

Second, hold the item and the records. Do not discard packaging, do not repack, do not add missing units, do not return anything to saleable inventory. Photograph the parcel as received before opening it further, including the label, the seal, and any damage. This hold applies until the case is either closed or explicitly released by the case owner.

Third, verify item and quantity against what was authorized: the approved SKU, the approved quantity, accessories, and serial numbers where they are recorded. Record what you actually received, not what you expected, and note the condition of each item.

Overhead hand-drawn illustration of a receiving bench organized as evidence, a sealed parcel photographed in place, a bench scale with a printed weight slip, a torn box flap held open with two fingers, a tablet showing a return authorization, packing tape roll set apart, muted palette with one warm accent, uneven hand-inked edges and slightly off-grid arrangement.
What step two produces: the parcel as received, not as tidied up

Fourth, align the handover timeline: carrier delivery, warehouse receipt, internal transfer, the moment of opening, and any inventory transaction. Discrepancies later in the case are usually a timeline gap, not a missing item, so build the sequence while memories and logs are fresh.

Fifth, test which path the evidence supports. That may point to one of the five paths described further down, or to nothing at all until more evidence arrives. Sixth, assign a case owner and a next update date. Unowned cases are the ones that age into write-offs.

The sequence can be drawn as a flow:

Flowchart: confirm parcel identity, hold item and records, verify item and quantity, align handover timeline, test which path the evidence supports, then assign a case owner and a next update date
The six-step intake sequence: each link protects the evidence the next one needs

Two constraints keep this sequence honest. The outcome may legitimately be undetermined, and when it is, the case must carry an explicit review or update date rather than an open status with no clock. And the hold must be released on a documented decision, including the decision to pay the refund without attribution, which is a valid business choice and should not be dressed up as a finding.

The evidence variables you must capture

Seven variables cover almost every empty-box and wrong-item dispute. Capture them at intake, in the same format every time, because a case is only comparable to your own history if the fields are consistent.

Variable What to record
Identity linkOriginal order, customer or tenant, RMA, parcel identifier, tracking number, receiving record
Expected goodsApproved SKU, quantity, accessories, serial numbers where available
Parcel conditionOuter packaging, seal type and integrity, labels, damage or repackaging traces
Weight evidenceSource of the measurement, time taken, device used, and whether the figure is billed weight
Inspection recordItems actually received, quantity, photographs, inspector, review outcome
Handover sequenceDelivery scan at the dock, warehouse receipt, internal transfer, opening, inventory transaction
Handling constraintsRefund status, channel rules, claim conditions, and the deadlines that apply

The parcel condition row is the one teams treat as housekeeping and later regret. Retaining packaging is not a formality. USPS states that for the claims it handles, the original packaging and contents must be kept available and may be requested for inspection (USPS domestic claims basics), which means a well-intentioned cleanup on the receiving bench can end a claim before it starts. The same instinct applies to repackaging a damaged carton for shelf space, which erases the cushioning pattern that distinguishes crush damage from a parcel that was packed without protection.

The weight row deserves its own discipline. Record who measured, when, and on what, because a weight with no provenance is an opinion. If your receiving scale is in a different unit or a different tolerance than the shipper's, note that too, and do not average the two as if they were the same measurement.

From finding to owner: the decision framework

Once the variables are captured, the intake finding usually falls into one of five rows. Each row names the next action and the party who owns it, which is what keeps a discrepancy from sitting in a shared queue.

Intake finding Next step Primary owner
Parcel or RMA link is wrongCorrect the mapping and re-verify the physical item3PL or systems owner
Outer packaging damaged or transport anomaly indicatedPreserve packaging, open an investigation or claim under the specific service usedShipping account holder with the carrier
Item identity or quantity mismatch, with inspection evidenceManual review, resolve the refund dispute under the applicable policyBrand returns owner with the 3PL
Warehouse records incomplete, or mismatch appeared after receiptInvestigate internal movement and inventory recordsWarehouse operations
Evidence insufficientGather more, set a handling deadline, do not attribute faultAssigned case owner

The table is used in one direction: start from the finding, land on the row, then land on a person. Two rows are commonly misapplied. The damaged-packaging row is owned by whichever party holds the carrier account, which is not always the warehouse that received the box, and the timing of the claim depends on the service used rather than on a single house rule. The insufficient-evidence row is not a verdict of no, and treating it as one pushes cases into refund refusals that later get reversed with interest in the form of a chargeback.

What the evidence can support: five paths

Five paths cover the outcomes that evidence can actually support. They are not mutually exclusive in the first hours, and a case may move between them as records arrive.

Path one, empty box or missing units. The question is whether the parcel was closed without the goods, or closed with them and lost them in transit. Ask for the packing or dispatch weight, any recording of the sealing step, and the receiving count. Where the outbound weight matches a full kit and the delivered weight matches an empty carton, the balance of evidence points to transit loss, and the path becomes a carrier claim. Where no outbound weight exists, the honest answer is that the case cannot be attributed, and the review step decides whether the commercial cost of a refund is lower than the cost of pursuing it.

Path two, the customer returned an item that is not this order's item. Compare SKU, serial or batch, and the mapping between the returned parcel and the original order. Give the customer a chance to self-correct first. Mis-shipped returns are frequently a second label in the same household or a mixed-up drop-off, and a short message with a photograph resolves more of these than any escalation.

Path three, the warehouse shipped the wrong item. This is the direction where the burden sits with the brand and the 3PL, and the customer is the injured party. Pull the pick records, the packing station photographs if you take them, and the same-batch impact to see whether other orders were affected. "The customer did not provide enough evidence" is not an available answer here, because the customer has nothing to prove about a parcel they never packed.

Path four, damage or pilferage in transit. Examine the outer carton, the cushioning pattern, and whether the packaging met the carrier's requirements. The customer refund line and the carrier claim line run in parallel rather than in sequence, for reasons covered in the next section.

Path five, contents repacked or substituted. This path is entered only when evidence points to it, never as a starting assumption. The indicators are specific: a weight that matches the expected contents while the received goods do not, a seal or tape type that differs from your standard, and serial numbers that resolve to other orders. Even here, the response is graded. A first occurrence with a cooperative customer usually ends in a refund with a note, while a pattern across addresses or accounts justifies a different conversation.

Two paths may hold at once, and a case may hold none. Both states are legitimate outputs of an intake process, provided the case carries an owner and a date.

Two clocks: customer refund vs carrier claim

The refund decision and the carrier claim are two separate processes with two separate clocks, and the most expensive mistake in this whole area is making the first wait for the second. A customer waiting on a carrier investigation will often escalate to a chargeback, which adds fees and a harder dispute to a case that was merely slow.

The claim windows are shorter than most teams assume, and their start dates are not the day anyone noticed the problem. FedEx requires a claim for damage or shortage within 60 days of the ship date (FedEx claims). UPS requires a claim within 60 days of the scheduled delivery date (UPS claims). USPS requires claims for damaged or missing contents on insured mail within 60 days of the mailing date (USPS claims). Card chargebacks run on a longer window measured from the transaction or statement, commonly around 120 days (chargeback window).

Bar chart of claim filing windows in days: a FedEx claim 60 days from ship date, a UPS claim 60 days from scheduled delivery, USPS insured mail 60 days from mailing date, and a card chargeback roughly 120 days from statement
Four windows, four different start dates, none of them the day the problem was noticed

The operational meaning of the two clocks is simple to state and easy to violate. The refund is decided on the evidence available at the time and on the commercial relationship with the customer, and it is booked internally. The carrier claim is pursued separately, with its own owner and its own recovery recorded against the same case, so that a successful claim later shows up as recovered cost rather than as the reason a refund was delayed. The date the discrepancy is filed is therefore a control point, not paperwork: file late on the outbound side and the claim window closes while the case is still open internally.

Retention, ownership, and the quarterly review

The chain runs from application to archive, and every segment needs an owner and a retention period. The table below is the working version of that contract, and the periods are planning defaults that you should adjust for your own category and claim conditions.

Stage Evidence to keep Owner Retention
Return requestReason code, declared SKU and quantity, customer photographs or video, request timestampBrand returns team12 months
Packing and handoverPacking or dispatch weight, seal method, carrier acceptance scanBrand or fulfillment site12 months, or through the claim window
TransitTracking history, delivery confirmation, delivery or billed weight, damage notesShipping account holderUntil the claim window closes and recovery is settled
Receiving and openingReceiving weight, unopened and opened photographs, opening video, inspection record, discrepancy report3PL12 months
Decision and settlementRefund or refusal decision, rationale, claim reference, customer communicationBrand returns owner24 months
Archive and reviewCase file, quarterly metrics, contract versionsBoth parties24 months

Naming and format rules matter as much as the retention periods. Keep original files rather than recompressed copies, name by case identifier and timestamp so a folder sorts chronologically, and set a defined retention for opening video, which is the single most useful artifact in an empty-box dispute and the first thing deleted when storage is tight. Compressing the photograph that shows a torn seal removes exactly the detail that makes it evidence.

A quarterly review of five numbers will tell you whether the process is working: empty-box rate, wrong-item rate, value in disputed refunds, carrier claim win rate, and the share of cases closed as undetermined. That last figure is the honest one. A queue with many undetermined cases is not necessarily broken, but a queue where those cases never close is a sign the ownership rule is being skipped.

Three clauses in a 3PL agreement or SLA do most of this work. A discrepancy filing deadline, expressed in hours from receipt, since the carrier clock is measured from an earlier date. A duty to retain opening video and packaging for a stated period. And an allocation of the burden of proof, including the conditions under which it reverses, such as a 3PL-held account or a warehouse-record gap.

None of this makes disputes disappear. It changes what a dispute costs, and it makes one outcome possible that is nearly impossible today: closing a case as undetermined, with a date and a signature, instead of leaving it open forever and paying for it twice.

Turn an anomalous return into a case with an owner

EasyShippingX is a channel-neutral shipping layer: it compares US carrier options in one place and executes them through the Portal, Excel, or the API. A returned parcel and the next outbound order run on the same quoting logic, so both ends of the trip can be checked with the same inputs, and the start date of a claim window has a reliable anchor.

See the Rate API Talk to Sales