Failed Delivery Rate

The failed delivery rate is the percentage of shipments that are not completed on the scheduled delivery attempt, calculated over all attempts in a period. It counts how many packages come back undelivered: the everyday analogy is the mail carrier who knocks on the door and finds nobody home, so the package goes back and the whole loop starts over. It is an exception metric, not a volume metric: the lower it is, the more efficient and profitable the last-mile operation.

It is not just a customer service number, it is a gauge of hidden costs. Every failure forces a second trip, burns fuel and driver time, and adds temporary storage, complaint handling and, in many cases, reverse logistics when the package is returned.

It should not be confused with the messaging delivery rate, which in the WhatsApp world measures message deliverability, how many messages reach the device. Here we are talking about physical packages that do not reach their destination: they share the word delivery, but they belong to different domains, one is logistics and the other is messaging.

How it is calculated

The formula is simple: divide the number of uncompleted deliveries by the total number of delivery attempts in a given period, and express the result as a percentage. If 12 out of 100 daily attempts come back undelivered, that day's rate is 12%. The key is defining the denominator correctly (attempts, not orders) and recording the reason for each failure, because without the cause on record the number describes the problem but does not help solve it.

Typical causes in Latin America

The reasons a shipment comes back tend to repeat, and almost all of them are avoidable:

• Informal or poorly geocoded addresses: references like "next to the corner store" that the system cannot pinpoint. Weak geocoding is one of the most frequent causes.
• Recipient not home: nobody is there to receive the package at the time of the visit.
• Restricted-access areas: gated communities, buildings without a doorman or remote rural areas that complicate the final stretch.
• Incomplete contact details: without a phone number to coordinate, the driver gets no second chance before heading back.

Why it matters to a midsize company in Latin America

Every failure has a real, cumulative cost: the second trip, the fuel, the driver's time, temporary storage, complaint handling and, when the package is returned, reverse logistics. A Loqate (2021) study puts the average cost of a failed delivery in the United States at US$17.20 per order, with bad address data among the most frequent and avoidable causes. That is why lowering the rate improves margin without selling a single extra unit. Much of the problem can be solved before dispatch (validating the address, confirming a time window and coordinating with the recipient by message), and that pre-delivery coordination is what a WhatsApp contact operation like Sellium handles.

How to lower the rate

Reducing it does not require adding trucks or switching carriers, but removing friction before dispatch. Three levers account for most of the improvement: validating and geocoding the address at the time of purchase, so the driver leaves with reliable coordinates; confirming a delivery window with the customer, so someone is there to receive it; and keeping a channel open to coordinate or reschedule if something changes. Every attempt resolved on the first visit is one fewer rework, and those savings go straight to profitability.

Difference from the messaging delivery rate

It shares a word with a term from the WhatsApp world, but it measures something completely different. It is worth being clear:

The failed delivery rate is pure logistics: it counts parcels that come back. The messaging delivery rate is deliverability. And OTIF looks at something else: whether the delivery arrived complete and on time, not just whether it arrived.

In practice

Picture an operation that ships 100 orders a day with 15% failing on the first attempt. Each retry adds a trip, so those 15 packages cost almost as much as if they were 30. By validating addresses at checkout and confirming the window by message, failures drop to 6: that is 9 fewer reworks per day, more than 200 a month, on exactly the same fleet. That is the kind of improvement that makes it tangible why the failed delivery rate is an exception metric worth watching closely.

Common mistakes

• Treating failure as inevitable instead of measuring and attacking its causes.
• Not validating or geocoding addresses before dispatch.
• Not recording the reason for each failed delivery, which makes it impossible to prioritize improvements.
• Confusing it with the messaging delivery rate and mixing two metrics from different worlds.