Last mile delivery is the final stage of a parcel's journey — from a local hub or depot to the recipient's door. It is typically the shortest leg by distance and the most expensive one per parcel, often by a wide margin.

That sounds like a contradiction until you look at what actually drives the cost.

Why distance stops mattering

For the long legs of a journey — city to city, hub to hub — cost scales with distance. A vehicle carrying a thousand parcels four hundred kilometres divides its fuel, driver and vehicle costs across all thousand. The per-parcel cost is small because the load is large.

In the last mile, that arithmetic inverts. The vehicle carries perhaps eighty parcels and has to stop eighty times. Each stop needs the driver to park, find the address, make contact, hand over the parcel, capture proof, and sometimes collect payment. That time does not reduce as the route gets shorter.

So the cost driver changes from kilometres travelled to stops made — and stops are far more expensive than kilometres.

Drop density: the number that matters most

Drop density is how many deliveries fall within a given area. It is the single largest influence on last-mile cost per parcel.

Consider two runs of sixty parcels. One covers a dense residential neighbourhood where stops are two minutes apart. The other is spread across a city, with fifteen minutes of driving between stops. Same parcel count, same working day — but the second run cannot physically be completed by one person, needs more vehicles, and costs several times as much per parcel.

This is why delivery businesses care so much about volume in specific areas rather than volume overall. Growth in a city you already serve densely is far more profitable than the same growth spread thinly across new ones.

Failed attempts: the hidden multiplier

A failed delivery does not cost a little extra. It roughly doubles the cost of that parcel, because the entire last mile has to happen again: same travel, same stop time, same handling. A parcel that fails twice has consumed three deliveries' worth of cost and produced one delivery's revenue — if it succeeds at all.

This is why first-attempt success rate is the metric most worth improving, and why the work that improves it happens before the van leaves: confirming the address, notifying the recipient, offering a window. See NDR vs RTO for how failures are handled once they occur.

Time at the door

The third driver is easy to underestimate. If proof capture, scanning and payment take ninety seconds instead of thirty, that is a minute lost per stop. Across sixty stops, that is an hour — which may be the difference between one run and needing a second person.

This is a software problem more than an operational one. A rider app that requires several screens, a stable connection, or careful typing at the doorstep is silently adding cost to every delivery in the network.

Why route optimisation helps less than expected

Route optimisation is the improvement everyone reaches for first, and it does help — sequencing stops sensibly saves real time. But it optimises the travel between stops, which is often not the dominant cost.

If your drop density is low, optimisation makes an expensive route slightly less expensive. If your first-attempt rate is poor, optimisation makes failed deliveries slightly more efficient. Neither addresses the actual problem. Density and attempt success are structural; routing is a refinement on top of them.

What actually reduces last-mile cost

• Cluster deliveries geographically — the largest single lever available
• Raise first-attempt success — address validation and pre-delivery notification
• Cut time per stop — fast scanning and proof capture, offline-first
• Resolve failures the same day — while a reattempt is still cheap
• Sequence routes properly — worthwhile, but after the above

What to measure

Total distance travelled is the metric most often reported and the least useful. These tell you more:

• Cost per successful delivery, by area
• Stops per hour
• First-attempt success rate
• Average time at stop
• Deliveries per square kilometre

Segment all of them by area. Aggregate numbers hide the thing you need to see, which is that some areas are profitable and others are not — and the average tells you nothing about which is which.

Where we come in

Delivery planning, drop clustering, rider applications and first-attempt improvement are what last mile delivery software and delivery management software are for. The field application side is covered in courier mobile app development.