LAST MILE · FULFILMENT · SORTING · ROUTING · DELIVERY · RECEIPT · EXCEPTIONS · RETURNS

How Last-Mile Delivery Works

Last-mile delivery is the part of logistics that converts a shipment inside a distribution network into a successful handoff at a specific real-world receiving point.

The last mile is where an optimised network meets an unoptimised street, building, doorway and human schedule.

An ocean container can cross half the world on a tightly scheduled vessel. A parcel can then lose a day because an address is incomplete, a lift is inaccessible, a recipient is absent or the route has too many stops for the available hours. Last-mile delivery is difficult because freight fragments from large consolidated flows into many small, unique handoffs.

The short answer

ORDER READY
  → SORT BY DESTINATION
  → ASSIGN DELIVERY DEPOT / ZONE
  → BUILD ROUTE
  → SEQUENCE STOPS
  → LOAD IN DELIVERY ORDER
  → DISPATCH
  → TRAVEL
  → LOCATE ADDRESS / ACCESS POINT
  → CONTACT / ACCESS
  → HAND OVER OR PLACE AS AUTHORISED
  → PROOF OF DELIVERY
  → EXCEPTION IF UNSUCCESSFUL
  → REDELIVERY / PICKUP / RETURN
  → DATA BACK TO PLANNING

1. The last mile begins before the driver leaves

Successful delivery depends on upstream fulfilment accuracy, address quality, service promise, route planning and load preparation. A driver cannot fix a parcel packed for the wrong customer or an address that never existed.

See How Warehousing Works for the order-fulfilment process that creates the delivery unit.

2. Delivery density changes economics

A vehicle serving many stops close together can deliver more orders per hour than one crossing long distances between sparse addresses. Density reduces travel between stops and lets fixed route cost be spread across more deliveries.

3. Geography is not only distance

Bridges, one-way streets, parking rules, gated estates, loading bays, islands, stairs, lifts and security desks all change effective travel and stop time. The straight-line distance between two customers may say little about the real route.

4. Address quality is infrastructure

Postal addresses, unit numbers, building names, access instructions and contact details must be precise enough for the delivery system to locate the receiver. Geocoding converts address information into usable route coordinates, but poor input can still map to the wrong entrance or building.

5. A route is a constrained sequence

Route planning asks which stops should be served by which vehicle and in what order. It may consider travel time, vehicle capacity, customer windows, service priorities, driver hours, access restrictions and expected stop duration.

The shortest geometric path is not necessarily the best operating route.

6. Time windows create scheduling commitments

A delivery promised between specific hours cannot be optimised like a route with no timing constraint. Each time window narrows sequencing choices and makes earlier delays propagate into later stops.

7. Stop time can matter more than drive time

Parking, finding the correct entrance, security registration, lift waiting, unloading, signature, payment or installation can make the time spent at a stop larger than the time spent travelling from the previous one.

Good route models therefore estimate service time as well as road time.

8. Loading sequence follows route sequence

If the first delivery is buried behind the last, every stop creates extra searching and handling. Vehicles should be loaded so the route can be executed without repeatedly rearranging cargo.

9. Sortation converts a parcel stream into routes

Distribution depots sort packages by postcode, zone, route, service level or delivery wave. Automated sorters can process high volumes, but the system still needs accurate labels and destination data.

10. Dispatch is a readiness gate

Before departure, the operator should know which parcels are on the vehicle, which route is assigned, whether special handling requirements are satisfied and whether any stop has changed.

A route leaving early but incomplete may create more rework than a route leaving correctly.

11. Traffic makes the route dynamic

Congestion, crashes, roadworks and weather change travel time after planning. Dynamic routing can update sequencing when new information materially improves the remaining route.

Too much re-optimisation can also confuse operations, so systems need stable decision rules.

12. The receiver is part of the route

A home, office, shop or construction site has its own receiving process. Someone may need to open a gate, provide identification, inspect goods, sign, operate unloading equipment or prepare cold storage.

Arrival is useful only when the receiving side can complete the handoff.

13. High-rise delivery adds vertical distance

Dense cities shorten road distance but can add lift queues, security desks, loading-bay rules and walking time inside buildings. Urban density improves some route economics while creating complex stop execution.

14. Offices and campuses create internal networks

The street address may be only the outer gate. Large workplaces, hospitals, schools and campuses can require internal delivery to departments, loading docks or mailrooms. The final hundred metres may have its own logistics system.

15. Failed delivery is a network multiplier

An unsuccessful attempt creates more than one failure. The parcel may need rehandling, storage, customer communication, route capacity for another day and possibly return transport.

Preventing one failed attempt can therefore remove several downstream tasks.

16. Delivery options trade convenience for complexity

Home delivery, scheduled windows, pickup points, parcel lockers and collection counters offer different combinations of customer convenience and operating density. Out-of-home delivery can consolidate many parcels into one stop while asking customers to complete the final movement themselves.

17. Lockers create a reusable receiving node

A parcel locker lets many deliveries terminate at one controlled location. It reduces dependence on recipient presence but introduces locker capacity, compartment sizing, access codes and collection deadlines.

18. Same-day delivery compresses every buffer

Same-day service requires inventory near demand, rapid order release, fast picking, frequent dispatch and enough local capacity. It is not merely a faster driver; the whole upstream network must be configured for short response.

This connects to How Logistics Network Planning Works.

19. Grocery delivery adds perishability and substitution

Fresh and chilled products impose condition requirements while customer orders may include many small lines. Short shelf life and item availability can create substitution decisions before the route even begins.

20. Cold-chain last mile preserves condition to receipt

Temperature-sensitive goods need suitable vehicles or packaging, controlled loading and rapid response to delays. The condition requirement remains active until the receiving point takes custody.

See How Cold Chain Logistics Works.

21. Bulky delivery is a different last mile

Furniture, appliances and industrial equipment may require two-person crews, lifting equipment, appointment windows, assembly or removal of old items. Stop duration and vehicle capacity dominate route design more than parcel count.

22. Proof of delivery closes the custody loop

Proof may include signature, photograph, timestamp, scan, recipient identity or location event depending on service and law. The purpose is to record that the promised handoff occurred under the agreed method.

23. Contactless delivery changes evidence, not responsibility

When direct signature is not required, the system still needs an authorised placement rule and enough evidence to resolve disputes or locate the item later.

24. Customer communication is an operating tool

Accurate estimated arrival windows, pre-arrival notifications and rescheduling options can improve first-attempt success. Communication is therefore part of capacity management because it changes receiver readiness.

25. Exceptions need coded reasons

“Failed delivery” is too vague for improvement. Wrong address, customer absent, access denied, damaged parcel, vehicle delay and unsafe conditions require different corrective actions.

Structured exception data turns operational failure into learnable information.

26. Returns make the last mile bidirectional

Drivers or pickup points may collect returns while delivering new orders. Combining forward and reverse flows can improve utilisation but adds identification, segregation and disposition requirements.

See How Reverse Logistics Works.

27. Peak seasons reshape the route network

Holiday demand and major promotions can multiply parcel volume over short periods. Operators may add temporary depots, vehicles, shifts, pickup points or delivery waves. Peak planning is capacity orchestration, not merely overtime.

28. Microhubs can move consolidation closer to customers

Urban microhubs receive consolidated inbound loads and redistribute them into smaller local delivery routes. They can support walking, cargo-bike or compact-vehicle operations where urban density and regulation make the model practical.

29. Electric vehicles change operating constraints

Electric delivery fleets can reduce local tailpipe emissions, but route planning must include vehicle range, payload, charging time and charging-site availability. The technology changes some constraints rather than removing planning.

30. Drones and robots solve narrow route cases

Autonomous delivery technologies can be useful in selected environments, but payload, range, weather, safety, regulation, access and economics constrain broad deployment. They should be evaluated against the actual stop problem rather than treated as universal replacements for conventional delivery.

31. AI can improve routing when the data describes reality

AI can estimate travel and stop times, predict failed deliveries and improve route sequencing. Poor address data, unrealistic service times or unrecorded access restrictions still create bad plans.

32. Last-mile metrics must include the customer outcome

Useful measures include first-attempt success, on-time delivery, stops per hour, distance per delivery, cost per stop, damage, complaints, redelivery rate and proof-of-delivery quality.

A route can look productive while creating poor receipt if speed is achieved through failed handoffs.

33. Common last-mile failure modes

Failure	Consequence
Bad address data	Routing begins from a false location.
Ignore stop time	Routes contain more work than the day can hold.
Poor load sequence	Drivers search and rearrange cargo at stops.
No receiver notification	First-attempt failures increase.
One failure code	The system cannot learn different causes.
Promise faster than network capability	Every downstream operation becomes an expedite.
No return path	Failed and unwanted deliveries become unmanaged stock.

34. The deeper model: the last mile converts a network destination into a human receipt

Upstream logistics can describe a destination as coordinates, a postcode or a route stop. The last mile must discover whether that abstract destination can actually receive the item now.

Delivery is complete only when the network’s idea of arrival becomes the receiver’s reality.

Continue the How Logistics Works series

• How Logistics Works
• How Warehousing Works
• How Logistics Network Planning Works
• How Cold Chain Logistics Works
• How Reverse Logistics Works

World Return: Follow one successful delivery backwards from the receiver’s hand. The route reveals address data, sorting, loading, capacity, inventory and network design. The last mile is where every upstream decision is finally tested by a real place and a real person.