What Does a Failed Delivery Actually Cost? The Bill Most Operators Never Add Up

A courier reaches a fourth-floor apartment, knocks, waits, calls the number on the label, waits again and leaves. A few minutes gone, one parcel back on the van. Ask most operators what that just cost and you get a quick answer: the price of sending the courier back. Fuel, a slice of driver time, the second attempt. That figure is real but it is nowhere near the total.

The rest of the bill is scattered across places the operations do not see. The van capacity that could have carried a paying parcel. The “where is my package” call answered on the customer-service budget. The returns team that processes it a week later. The shipper scorecard quietly marked down. Added together, a failed home delivery costs several times what the retry line suggests, but few operators ever run that sum, because no single person owns it.

This article follows that scattered bill to its total, in B2C e-commerce home delivery across Europe: what the real number is, why it stays hidden and which parts of it you can actually move.

Nobody is measuring the same thing

Before you can put a price on a failed delivery, you have to agree on what one is. Unfortunately, the industry has never settled that. The lack of a shared definition is the first reason the cost stays invisible.

The First Attempt Delivery Rate (FADR) is the metric everyone quotes. It’s also close to useless for comparison, because it can mean something completely different at every company that reports it:

• A parcel dropped into a locker counts as a successful first delivery, despite the fact it has not reached the recipient and may be shipped back unclaimed after 72 hours.
• A mid-route redirect to a PUDO point gets logged as a failed attempt at one operator and as a straight PUDO delivery, with no failure on record, at the next.
• In the UK, “carded” (a card left at the door) and “attempted” are two separate events. In some other markets, they are one.
• At some operators a second attempt opens a new record in the system, so the original failure quietly falls off the counter.
• A parcel marked as delivered that never reached the customer does not register as a failed delivery at all. It shows up as a complaint, in another department, on another report.

Put those together and you see the problem: a network with 60% out-of-home penetration can report 96% FADR and still lose money on the 40% it delivers to the door. Measured this way, FADR tells you how good your channel mix looks, not how well you deliver. A more honest measure is the number of touches it takes to get one parcel to its recipient.

Thus, for the rest of this article, we’ll stick to the following definition:

A failed delivery is any parcel that needs an extra touch after it has left the depot on a route.

It covers the recipient being out, no access to the building, a wrong address, a refused parcel, a failed cash-on-delivery payment, a mid-route redirect, a locker parcel left unclaimed inside its collection window and the “delivered” parcel the customer never got. It leaves out line-haul delays, missorts in the hub and damage, which are real problems that carry their own bills, just not this one.

The bill nobody adds up

Here is the full cost of one failed delivery, broken down by where each piece lives inside the company. That last column is the whole point.

Cost component	Where the cost sits
Second attempt: courier time, fuel, vehicle wear	Operations – the only line someone sometimes tracks
Re-handling at the depot: intake, re-sorting onto a route, storage	Operations – buried in general depot overhead
Route disruption and the occupied van capacity	Structurally unmeasurable
Customer contact after the failed attempt	Customer experience – a different budget from operations
Return to sender after repeated attempts, including returns sorting	Reverse logistics – visible as a returns cost, not a failed-delivery cost
Commercial fallout: shipper scorecard, contractual penalties, a lost tender	Sales – entirely outside the operations ledger

There is one more item that is only now becoming a cost: per-parcel emissions reporting under CSRD and Scope 3. Failed deliveries inflate the emissions per delivered parcel, and this is increasingly a contractual requirement from large shippers. Today, it is a reporting nuisance. Within a couple of years, it will carry a price.

Now, which of these is the biggest and the easiest to overlook? Two of them, each invisible for its own reason.

Route disruption and consumed capacity. A failed attempt does not cost you one stop. It costs more than that, in three layers:

1. Doorstep time – the courier loses time on the spot: calling, waiting, trying the intercom, deciding what to do next. That’s a few minutes, spent in the densest part of the route.
2. A costlier retry – the retry costs more than an average stop. That’s because the parcel returns to the depot and gets slotted into the next day’s route, one planned around a different set of addresses – usually as a geographic outlier.
3. Lost van capacity – every retry occupies a slot in the van that a new, revenue-earning parcel could have filled. Through November and December, that stops being an accounting cost and becomes a ceiling on throughput, and peak capacity is exactly what lets you take on a large new shipper. Failed deliveries quietly consume the very room you need to grow.

Customer contact. This one hides for a structural reason. The cost of the calls and messages a customer sends after a failed attempt lands in the customer experience budget, not operations. As a result, a COO looking at the cost of failed deliveries does not have that figure in their own P&L. Not for lack of wanting it. It simply lives one department over.

What it actually costs

Are there actual numbers to put on this? There are some specific results worth knowing, but all of them come with an asterisk:

• 14.69 EUR in Germany and 11.60 GBP in the UK – drawn from a 2020 Loqate/Censuswide study. The caveat is that wages, fuel and channel mix have all moved on since then.

The figure matters less than the mechanics behind it. According to The Last Mile Experts Out-of-Home Delivery in Europe 2025 Report, labor costs make up 60% of the total. The cost tracks local wages, not the value of the parcel, which is why one European number is close to meaningless. An operator in Bucharest and one in Oslo do not have the same problem measured in euros. When it comes to Central and Eastern Europe, the amount is lower, the proportions between the components hold.

Why the cost of a failed delivery stays hidden

You’d think that with a cost this large, somebody would calculate it. Almost nobody does and there are four reasons (listed from the weakest to the most important one):

1. The cost is scattered across departments. Depot, line-haul, customer service, complaints, sales, each with its own P&L. “Failed delivery” is not a cost object in anyone’s ERP. Nobody owns the total.
2. The reason-code data is not good enough to calculate from. A courier has a handful of reason codes in the app. But in practice, under time pressure, they keep using the same two. A large share of failed attempts gets logged as “recipient not home” regardless of what actually happened. No trustworthy reason codes mean no root-cause analysis which, in turn, means no case for change.
3. There is no tool for it. A transport system records events, not costs. Getting the number requires converting the event log into a cost, which needs a data warehouse and a cost-allocation work owned by finance, not operations.
4. A conscious choice. The operations director’s bonus hangs on cost per parcel and FADR. Showing the full cost of failed deliveries makes their own numbers look worse. Nobody volunteers to document that their department loses a few million euros a year.

The fourth reason is the one that matters most and it’s worth being honest and plain about it. The shortage of data is real, but it mostly serves as cover for an incentive that points the other way.

The causes of failed deliveries

Here is the rough breakdown of why first attempts fail, starting with the most common causes:

• Recipient not home
• Address data quality and geocoding
• No access to the building: intercom, gate code, security
• Bad route sequencing, arrival outside the window the recipient was in
• Refused parcel, cash-on-delivery payment problem

Two more belong on the list without a clean percentage: attempts that never actually took place (detectable by comparing GPS location and dwell time against the address, rarely measured), and attempts logged for a parcel that was not even on the van in the first place.

In addition, there are a couple of things that matter more than the ranking:

First, “recipient not home”. It’s worth pointing out that it’s not really a cause. In fact, it’s a result. The cause is that the recipient did not know when the courier would arrive. A “sometime today between 8 and 18” window is hardly what a customer can plan around. But give them a narrow window and a one-tap way to redirect the parcel and they behave completely differently. Once you relabel a large share of those failures from “out of our hands” to “our communication problem”, they from the list of things you complain about to the list of things you can act on and change.

And second, address data quality is underrated, badly. It shows up as wrong geocodes, no reference data for new estates, missing staircase and floor numbers, the same street name repeating across an agglomeration.

What software can fix and what it can’t

Can technology help with reducing failed deliveries? The thing is, the biggest lever here is not software. It’s a channel mix.

According to Last Mile Experts’ Out-of-Home Delivery in Europe 2025 Report, delivery to a locker or PUDO point has a first-attempt success rate above 99%. Their 2026 Ultimate Out-of-Home Delivery Handbook, also sets out the productivity gap: 100-150 parcels per route for home delivery against up to 1,400 out of home. No routing algorithm evens that out. But there are areas where technology genuinely comes in handy:

• A precise delivery window and estimated time of arrival (ETA), plus a one-action way to reschedule or redirect, available both before the route and during it.
• Address data quality treated as a continuous process rather than a one-off cleanup, including learning from the confirmed locations of past successful deliveries. Every successful delivery is a free, precise geocode for that address, and most operators discard it.
• The address held as a durable object with a history, such as gate codes, “intercom broken, call on arrival,” and the best time of day to try, rather than a text field printed on the parcel.
• Reason-code discipline: structured, mandatory and cheap enough for a courier to enter at the door.
• The next-step decision made at the door, not back at the depot the next morning, whether that is a redirect to the nearest point, a hand-off to a neighbor, or a retry.
• Real-time visibility, with FADR and reason codes split by region, route, and courier, with alerts.

Several of these gains depend on good routing and AI route optimization is now doing much of that work in the field.

Some causes stay out of technology’s reach entirely: cash on delivery, weather, gated estates, recipient who is simply not there and refuses out-of-home delivery. Others yield only halfway: building access, where data helps until physics takes over, and courier behavior, where software gives you visibility but the fix is a management job.

To put it simply, software does not make failed deliveries disappear. What it does do is route the parcel into the right channel and make sure it arrives when somebody is home to receive them. Purpose-built courier management software is what turns that from an aspiration into a repeatable process.

Questions worth asking your own team

The cost of a failed delivery is real, large and scattered, but there is no single button to press here that would fix all this. What there is, is a set of questions. If you run last-mile operations, take these to your team:

• What is our first-attempt success rate, split between home delivery and out-of-home? If we have one number for the whole operation, we know nothing.
• Where does that number come from (A real-time panel, or a monthly report?) and how old is it?
• How many reason codes do we have and how many do couriers actually use?
• Who in the company knows the total cost of failed deliveries, customer service and returns included? If the answer is nobody, you are not short of data. You are short of an owner.
• How many retries are taking up space in our vans in November?

If you have run these numbers and landed somewhere else, we would be glad to compare notes. Get in touch with our experts.

FAQ

Isn’t FADR still a useful metric?

As an internal trend line for a single, consistently measured operation, yes. As a number to compare against another operator or to report as a headline of delivery health, no, because the definition is not standardized and a high FADR can coexist with heavy losses on home delivery. Split it by channel before you trust it.

We already track the cost of the second attempt. Isn’t that the main cost?

It is the most visible cost, which is not the same thing. The second attempt is usually the smallest of the pieces once you add eaten van capacity, customer-service contacts sitting in another budget, returns handling, and commercial fallout on the shipper scorecard. Tracking only the retry is how the full bill stays hidden.

What is the fastest way to reduce failed deliveries?

For the share of volume that can go out of home, shifting it to lockers or PUDO points, because first-attempt success there is above 99%. For genuine home delivery, better arrival windows and a one-tap redirect option, which attack the largest cause (“recipient not home”) at its actual root, which is communication.

How do we even start calculating our own cost per failed delivery?

Start with ownership, not a spreadsheet. Name one person who is allowed to pull the pieces from operations, customer experience, reverse logistics, and sales into a single figure. The calculation is not hard once someone is permitted to cross those departmental lines. The reason it has never been done is that nobody was.