Every retailer with a shipping operation eventually meets the same quiet crisis: a tracking page that stops moving. The last scan is three days old, the carrier status reads “exception,” and the customer has already emailed twice. Nobody has stolen the parcel and nothing has gone catastrophically wrong. Something simply broke in the handoff chain, and the clock is now running against both the order and the margin on it.
Delivery exception handling is the discipline of deciding what that silence means and what to do about it. Done badly, it produces a support queue full of “where is my order” tickets, duplicate shipments sent to customers who eventually receive both parcels, and carrier claims filed after the filing window closed. Done well, it is close to invisible: most exceptions resolve themselves, the handful that will not are caught early, and the customer hears from you before they think to ask.
In short
- Most exceptions clear without intervention. Weather holds, missed connections and sort-facility delays usually resolve in 24 to 72 hours, and acting on day one wastes money.
- The scan gap matters more than the status text. A parcel with no new scan for three business days is a different problem from one showing a fresh “delayed” code at a sort hub.
- Address failures are the one exception type that never self-resolves. They need a correction submitted through the carrier, usually within a short holding window, or the parcel returns to sender.
- Reship, refund and claim are three separate decisions. Deciding them together leads to reshipping orders you should have refunded and claiming on parcels the carrier will never pay out on.
- Claims fail on documentation, not on merit. Carriers publish filing windows and evidence requirements; missing either one closes the case regardless of what actually happened to the parcel.
What a delivery exception actually is
A delivery exception is a carrier-generated flag meaning the parcel has deviated from its planned path or timeline. It is a status, not a verdict. The carrier is telling you that its own routing plan no longer matches reality and that the original delivery estimate should not be trusted.
That definition is broader than most retailers assume. An exception can mean a truck was delayed by snow, a barcode would not scan, a customs form was incomplete, a gate code was missing, or a recipient was not home for the third attempt. These sit at wildly different points on the severity scale, and the tracking page rarely makes the difference obvious.
The practical consequence is that “exception” on its own is not an actionable signal. What matters is the underlying reason code, how long the parcel has been static, and where in the network it stopped. A parcel flagged at an origin facility one hour after pickup is a scanning artifact. The same flag on a parcel that has not moved out of a regional hub for four days is a real operational problem.
Retailers who are new to this layer of the operation often benefit from understanding the broader delivery chain first, because exceptions are symptoms of handoffs. Our primer on last-mile delivery for retailers who never thought about it covers how parcels actually move between the injection point, the sort network and the final courier, and that structure explains where most exceptions are generated. The wider context sits in our guide to modern retail logistics from warehouse to doorstep, which maps the same chain end to end.
Exceptions versus delays versus losses
Three terms get used interchangeably and should not be. A delay means the parcel is moving but will arrive later than promised. An exception means the carrier has flagged an obstacle. A loss means the carrier can no longer locate the item at all.
Only the third category triggers a claim. The first two are customer-communication problems that become claim candidates only if the parcel never reappears. Teams that collapse the three into one bucket end up filing claims on parcels that are merely late, which burns goodwill with carrier account managers and clutters the claim record you will want clean when a genuine loss happens.
The exception codes that matter and what they mean
Carriers publish their own status vocabularies, and they do not align. What UPS calls an “address correction required” event, USPS may describe as “insufficient address,” and a regional courier may surface as a generic hold. Exception code definitions and their handling requirements are published by each carrier in its own service guide or terms, and those documents are the authoritative reference rather than any third-party summary, including this one.
Despite the naming differences, exception reasons cluster into a small number of operational families. Sorting incoming flags into those families is more useful than memorizing per-carrier code lists, because the correct response depends on the family rather than the wording.
| Exception family | What it usually means | Typical self-resolution | First action |
|---|---|---|---|
| Weather or natural event | Network-wide hold in a region, not parcel-specific | 24 to 72 hours, often longer in peak season | Wait. Check the carrier service alert page for the affected region. |
| Network or sort delay | Missed connection, volume surge, facility backlog | 24 to 48 hours | Wait one business day, then check for a new scan. |
| Address issue | Incomplete, undeliverable or mismatched address data | Does not self-resolve | Submit a correction immediately through the carrier portal. |
| Failed delivery attempt | No safe location, signature required, access blocked | Retries on following business days, then returns | Contact the customer with pickup or redirect options. |
| Customs or documentation hold | Missing paperwork, duty or classification query | Days to weeks, highly variable | Supply the requested documentation to the broker or carrier. |
| Damage or repack | Packaging failure detected in the network | Varies; may continue or be returned | Prepare claim documentation. Do not wait for delivery confirmation. |
| No scan at all | Label created, never inducted, or scan skipped | Ambiguous; often a missed scan rather than a missing parcel | Verify the parcel physically left your facility. |
The label-created trap
The single most misread status is “label created” or “shipment information received” with no movement afterward. Customers see it as proof the order was never shipped. In practice it means only that a label was generated and the carrier has not yet recorded an induction scan.
Roughly half of these cases are scanning gaps where the parcel is genuinely moving but unscanned until it reaches a sort hub. The other half are real: the parcel is sitting on a bench in your own warehouse. Before you escalate anything to the carrier, confirm the outbound manifest shows the item left the building, because a carrier investigation on a parcel still in your building wastes several days.
Why code mapping is worth building once
If you ship across more than one carrier, build a normalization table that maps each carrier’s raw status strings into your own internal families. Every downstream decision, the waiting rules, the customer message templates, the claim triggers, then runs off your vocabulary instead of theirs.
This is a one-time piece of work that pays out permanently. It also makes carrier comparison possible, because you cannot compare exception rates across carriers whose status taxonomies do not match. Our comparison of same-day delivery economics for retailers shows how sensitive service-level decisions are to that kind of normalized measurement.
How long to wait before acting on a stalled scan
The expensive mistake in delivery exception handling is acting too early. Reshipping on day two produces a measurable rate of double deliveries, where the original parcel surfaces a week later and the customer keeps both. Acting too late is also expensive, but in a different currency: refund requests, chargebacks and negative reviews.
The resolution is a waiting rule keyed to the scan gap rather than to the exception status. A scan gap is the elapsed business time since the last tracking event of any kind. It is the cleanest available proxy for whether the parcel is still in the network.
A workable waiting ladder
Most retailers converge on something close to the following thresholds for domestic ground shipments. Adjust them upward for peak season, and downward for express services where the promise is tighter.
- Scan gap under 2 business days: no action. Suppress the internal alert entirely and let the network work.
- Scan gap of 2 to 3 business days: proactive customer message. No carrier contact yet, no reship.
- Scan gap of 3 to 5 business days: open a carrier trace or investigation. Tell the customer you have done so, and give a date by which you will resolve it either way.
- Scan gap over 5 business days: treat as presumed lost. Move to the reship or refund decision without waiting for the carrier to conclude its trace.
Two qualifiers make this ladder work in practice. First, business days exclude carrier non-operating days, which matters enormously around public holidays when a 3-day gap can be entirely normal. Second, express and same-day services should compress every threshold, because a two-day silence on a next-day service is already a service failure.
Do not wait on address and customs exceptions
The waiting ladder applies to silence, not to flags that explicitly request something from you. Address corrections, customs documentation requests and “action required” holds all carry their own countdown clocks, and the parcel typically returns to sender if you miss them. These get worked the same business day they appear, regardless of scan gap.
That distinction is worth encoding directly in your alerting. Silence-based exceptions go into a queue reviewed daily. Action-required exceptions go into a queue reviewed hourly. Mixing them means the urgent ones sit behind a hundred weather delays that need nothing from anybody.
Failed delivery attempts and address correction
Failed attempts and address problems together account for a large share of exceptions that never resolve on their own, and they are the ones most within a retailer’s control. Both trace back to data captured at checkout, which is where the cheapest fix lives.
A failed attempt means the courier reached the address and could not complete delivery. The usual causes are a signature requirement with nobody present, a secured building with no access, no safe drop location, or an aggressive dog. Carriers normally retry on the next business days and then hold the parcel at a facility for a limited period before returning it.
An address exception means the courier could not reach the address at all: a missing apartment number, a transposed street number, a ZIP code that does not match the city, or a commercial address delivered on a weekend. These require a correction, and corrections are usually chargeable. The fee is small relative to a return-to-sender cycle, so approving it quickly is almost always the right call.
Fixing the source rather than the symptom
Retailers who see a persistent address-exception rate above roughly one percent of shipments usually have a checkout problem rather than a carrier problem. Address autocomplete, real-time validation against a postal database, and a secondary address line that is actually required for known multi-unit buildings remove most of it. We covered that economics in detail in our piece on address errors at checkout and the delivery costs they create.
The other structural fix is giving customers a collection option. Offering pickup at a store or locker converts a category of failed home deliveries into a completed order, and it is the same infrastructure behind buy online pick up in store. Where a store network exists, redirecting a stalled parcel to a nearby collection point is often faster and cheaper than a second delivery attempt.
What to tell the customer during a failed attempt
Carrier notifications on failed attempts are famously unhelpful, often a door tag with a reference number and no context. Your message should do the work the carrier notification does not: state plainly that an attempt was made, give the specific options available (redeliver, redirect to a pickup point, change the address), and give the deadline before the parcel returns.
Including the deadline is the part most retailers skip and the part that changes behavior. A customer who knows the parcel returns on Friday acts before Friday. A customer told only that “delivery was attempted” does nothing.
Deciding between reship, refund and claim
Once a parcel is presumed lost, three decisions follow and they are independent. Whether to send a replacement, whether to refund, and whether to pursue the carrier for the value are separate questions with separate answers. Treating them as one question is the most common structural error in delivery exception handling.
The reship-or-refund decision is about the customer relationship and inventory position. The claim decision is about recovering cost from the carrier and has no bearing on what the customer receives. You should resolve the customer within days and pursue the claim over weeks, in parallel.
| Situation | Customer resolution | Pursue carrier claim? | Reasoning |
|---|---|---|---|
| Low-value item, in stock, loyal customer | Reship immediately | Usually not worth the admin time | Claim handling cost can exceed the item value |
| High-value item, in stock | Reship after basic verification | Yes, with full documentation | Recovery materially affects margin |
| Item out of stock, no restock date | Refund promptly | Yes if declared value or insurance applies | Holding a customer for unknown weeks damages retention |
| Repeat loss claims from same account | Refund once, then require signature | Case by case | Pattern may indicate friendly fraud rather than carrier loss |
| Damaged on arrival, item replaceable | Reship and arrange disposal | Yes, photos are essential | Damage claims have the strongest evidence trail |
| International parcel held in customs | Wait, then refund if abandoned | Rarely payable | Customs holds are generally excluded from carrier liability |
Where returnless resolution fits
When a damaged parcel does arrive, the instinct to demand the item back is often uneconomic. If the return shipping plus processing exceeds the salvage value, writing it off is the cheaper path, and the customer experience is markedly better. The same arithmetic that drives returnless refunds applies directly to damage exceptions.
Set the threshold explicitly rather than case by case. A published internal rule (“items under a stated value are written off without return”) lets support agents resolve in one message instead of escalating, which is where the real cost of exceptions accumulates.
Guarding against pattern abuse
A minority of “lost parcel” reports are not carrier failures. Delivered-not-received claims cluster around certain addresses and certain payment profiles, and a policy of automatic reshipment invites repetition. The defence is not suspicion of every customer but a tracked history: flag accounts with multiple non-delivery claims, and require a signature service on their subsequent orders.
Keep the tone neutral in any customer-facing message. You are describing a shipping-method change, not making an accusation, and framing it otherwise creates a support escalation worse than the original loss.
Filing carrier claims that actually get paid
Carrier claims are rejected far more often for procedure than for substance. The parcel really was lost, the value really was what you said, and the claim is still denied because it arrived three days outside the filing window or without the commercial invoice attached. Treating the claim as a documentation exercise rather than an argument changes the outcome rate substantially.
Filing windows, liability limits, documentation requirements and the list of excluded goods are set by each carrier in its published tariff, service guide or terms of carriage, and they differ by carrier, by service level and by domestic versus international movement. Those published documents are the only reliable source, and they are revised periodically. Confirm the current figures with the carrier before relying on any timeframe, including the general patterns described below.
The documentation set to assemble before filing
Assemble the file before you open the claim rather than after. Most carrier portals time out a partially completed claim, and chasing a warehouse photo two weeks later rarely succeeds.
- Tracking number and full scan history exported at the time of filing, because carrier tracking data is not retained indefinitely.
- Proof of value: the commercial invoice or, for own-manufactured goods, a cost breakdown. Retail price is frequently not the payable amount.
- Proof of shipment: the manifest or pickup record showing the parcel entered the network.
- Photographs of the packaging and the damaged contents for damage claims, including the shipping label and any carrier-applied repack tape.
- Customer statement confirming non-receipt, dated, for loss claims where a delivery scan exists.
- Packaging specification for damage claims, since carriers routinely deny where packaging is judged inadequate for the contents.
Declared value is not insurance
Carriers generally state in their own terms that declared value represents the maximum limit of their liability and is not an insurance product. The practical effect is that declaring a higher value raises the ceiling on a payout but does not guarantee one, and the carrier can still deny on packaging, on an exclusion, or on the goods being outside its accepted list.
High-value or frequently lost categories are often better covered by third-party shipping insurance or by self-insuring through a reserve, since the premium structure differs from carrier declared value. Whichever route you choose, price it against your actual loss rate rather than against a worst case, and revisit it annually.
Freight moves under different rules
Parcel and freight claims are not the same regime. In the United States, interstate motor carrier liability for freight is governed by the Carmack Amendment, codified at 49 U.S.C. § 14706, which sets its own minimum periods for filing claims and bringing suit. If any part of your inbound or outbound flow moves as LTL or truckload, the applicable rules, timeframes and liability limits differ from the parcel terms your ecommerce team is used to. A general overview of the statute is available on Wikipedia, though the statute text and your carrier’s bill of lading terms are the operative sources.
Tracking claim performance
Log every claim with its filing date, amount claimed, amount paid and resolution date. Two numbers fall out of that log and both are negotiating leverage: your recovery rate (paid divided by claimed) and your average days to settlement. Carrier account managers respond to a documented pattern of slow or denied settlements in a way they do not respond to anecdotes.
The same log tells you when claiming is not worth doing. If small-value claims recover at a low rate and consume meaningful agent time, setting a floor below which you simply absorb the loss is a rational policy rather than an admission of defeat.
Proactive messaging that stops the support ticket
The cost of an exception is rarely the parcel. It is the two or three support contacts the exception generates, each with an agent cost and each arriving with the customer already annoyed. Messaging first converts that interaction from a complaint into an acknowledgement.
The timing rule is simple: the customer should hear about the problem from you before they check tracking themselves. In practice that means firing a message when the scan gap crosses your two-day threshold, not when the carrier finally updates the status days later.
What a good exception message contains
Four elements do almost all the work, and anything beyond them reduces clarity.
- What happened, in plain language. “Your parcel has been delayed in the carrier’s network” beats a pasted exception code every time.
- What you have already done. Naming the concrete action (a trace opened, a correction submitted) removes the customer’s need to chase.
- What happens next and when. A specific date for the next update, even if that update may be “still no news,” prevents the follow-up email.
- What the customer can choose. Redirect, pickup, replacement or refund, with any deadline stated explicitly.
What to avoid saying
Two habits generate escalations. The first is blaming the carrier by name, which invites the customer to conclude you chose a bad carrier and reads as evasion. The second is promising a delivery date the carrier has not committed to, which converts one disappointment into two.
Commit instead to your own next action, which you control. “We will update you by Thursday” is a promise you can keep even when the parcel does not move, and keeping it is what preserves the relationship. Reliability of communication substitutes surprisingly well for reliability of delivery, a pattern explored further in our analysis of delivery promise reliability and repeat purchase.
Tracking exception rates by carrier and lane
Individual exceptions are operational noise. Exception rates are a management signal, and they only become visible once you measure them consistently. The unit of analysis that matters is the lane, meaning an origin facility paired with a destination region and a service level.
Aggregate exception rates hide almost everything useful. A carrier running at a respectable overall rate may be generating most of its exceptions in one metropolitan area or on one service tier, and that pattern is actionable in a way the headline number is not.
The metrics worth maintaining
| Metric | Definition | Why it matters | Review cadence |
|---|---|---|---|
| Exception rate | Shipments with at least one exception, divided by total shipments | Headline health indicator per carrier | Weekly |
| Unresolved exception rate | Exceptions still open after five business days | Isolates real failures from network noise | Weekly |
| Address exception rate | Address-family exceptions divided by shipments | Measures your checkout data quality, not the carrier | Monthly |
| Mean scan gap at alert | Average days of silence when your alert fired | Tests whether thresholds are tuned correctly | Monthly |
| Cost per exception | Support time, reship cost and refunds, divided by exceptions | Turns a service metric into a budget line | Quarterly |
| Claim recovery rate | Amount recovered divided by amount claimed | Direct measure of claim process quality | Quarterly |
Using the data in carrier negotiations
Carrier pricing conversations are usually framed around rates and surcharges. Exception data lets you reframe them around total cost, which is where a retailer has more leverage. A carrier whose published rate is lower but whose unresolved exception rate is double a competitor’s is not cheaper once support, reship and refund costs are attributed to it.
Multi-carrier routing as the endgame
The mature version of this capability is routing rather than reporting. Once lane-level exception data is trustworthy, the shipping rules engine can steer parcels away from the specific lanes where a carrier underperforms while keeping it on the lanes where it is strong.
This is a meaningful engineering commitment and it is not where a small operation should start. The sequence that works is measure, then renegotiate, then route. Skipping to routing without reliable measurement produces rules nobody can justify six months later.
Important note on scope
This article is general information about operational practice and is not legal, insurance, customs or tax advice, and it does not create any professional relationship. Carrier liability, claim eligibility, filing deadlines and customs obligations depend on the specific contract of carriage, the service purchased, the goods involved and the jurisdictions concerned. For your own situation, consult a licensed customs broker, a freight or trade attorney, or your insurance adviser, and verify any timeframe or liability figure directly with the carrier’s current published terms.
FAQ on delivery exceptions
Does a delivery exception mean my parcel is lost?
No. An exception means the carrier has flagged a deviation from the planned route or schedule. The large majority resolve within 24 to 72 hours without anyone doing anything. Loss is a separate status that applies only when the carrier can no longer locate the item, and most exceptions never reach it.
How long should a retailer wait before reshipping a stalled parcel?
A common working rule is to take no action under a two-business-day scan gap, message the customer between two and three days, open a carrier trace between three and five days, and treat the parcel as presumed lost beyond five business days. Compress every threshold for express services, and extend them during declared weather events and peak season.
What is the difference between a delay and an exception?
A delay means the parcel is still moving but will arrive later than promised. An exception means the carrier has recorded a specific obstacle, which may or may not affect the arrival date. Exceptions requiring action from you, such as address corrections, are the ones that do not resolve on their own.
Who pays for an address correction fee?
Carriers typically bill the correction fee to the shipper account, meaning the retailer, regardless of who made the error. Some retailers pass it on for clearly customer-entered errors, though most absorb it because the dispute costs more than the fee. Reducing the underlying rate through address validation at checkout is the more durable fix.
How long do I have to file a carrier claim?
It varies by carrier, service and shipment type, and each carrier publishes its own filing windows in its tariff or service guide. Those windows are measured in weeks rather than months for many parcel services, and late filing is a standard denial reason. Check the current window in the carrier’s published terms as soon as a loss becomes likely rather than when you are ready to file.
Will a carrier pay out the retail price of a lost item?
Usually not. Carriers generally settle on documented cost or declared value, whichever is lower, subject to their liability limits and exclusions. Retail price, lost margin and shipping cost are commonly excluded, so the payout rarely makes the shipment whole.
Can I claim on a parcel held in customs?
Generally no. Carrier terms typically exclude delays caused by customs, regulatory inspection or incomplete documentation from their service commitments and liability. The practical response is to supply the requested documentation quickly, since most customs holds clear once the paperwork is complete. Rules and procedures differ by country, so verify with the relevant customs authority or a licensed broker.
What exception rate is normal for an ecommerce retailer?
There is no universal benchmark, because the figure depends heavily on service mix, geography, season and how each carrier defines an exception. The useful comparison is against your own trend and across your own carriers on matched lanes. A rising unresolved exception rate matters far more than the absolute level.
Should exception handling sit with support or with logistics?
Detection and resolution belong with logistics, because the decisions are operational and involve carrier relationships. Customer communication belongs with support, working from templates logistics maintains. Splitting it the other way produces agents guessing at carrier behaviour and shipping teams writing customer messages.
Exceptions are a permanent feature of parcel networks rather than a problem to be eliminated. The retailers who handle them well are not the ones with the lowest exception rates; they are the ones who classify quickly, wait the right number of days, communicate before being asked, and keep a clean enough claim record to recover what is recoverable. For the wider operational picture that this sits inside, our guide to modern retail logistics from warehouse to doorstep covers how the upstream decisions shape how many exceptions you see at all.