Skip to content
MGBack
Operations

Routing 500 daily deliveries without a routing team

The constraint model behind next-day delivery at national scale.

ON THIS PAGE

Next-day delivery is not a routing problem first. It is a promise problem: you decide what you are willing to promise, and routing is what makes the promise survive contact with traffic.

The constraint nobody writes down

Every routing conversation starts with distance and vehicle capacity. The constraint that actually broke our plans was the courier's own knowledge of a district — a route that is optimal on a map is slow for someone who has never worked that neighbourhood.

  • Courier familiarity with the district
  • Building access windows, not just delivery windows
  • Recipient availability by time of day
  • Vehicle capacity, last and least binding
We stopped optimising routes and started optimising who gets which district.

Sequencing beats optimising

Holding couriers on stable districts and sequencing within them beat a globally optimal daily reassignment by a wide margin, because familiarity compounds and reassignment resets it every morning.

Daily re-optimised5.4
Weekly districts8.2
Stable districts9.9
Stops completed per courier-hour, by assignment strategy

The lesson generalises past logistics: when a system has a human in the loop, the human's accumulated context is an asset on the balance sheet, and reshuffling assignments writes it off every cycle.

FOOTNOTES

[1]Figures are indexed within a single city operation over eight weeks; absolute rates differ sharply between cities.