Clarke and Wright's savings trick made delivery routes computable by hand

In 1964, G. Clarke and J. W. Wright, working with the Cooperative Wholesale Society in Manchester, published 'Scheduling of Vehicles from a Central Depot to a Number of Delivery Points' in Operations Research. Their target was the routing problem Dantzig and Ramser had formalized in 1959 — and for which exact solutions were too slow to compute.

Their savings algorithm is almost embarrassingly simple: start with a separate round trip for every customer, then for each pair of customers compute how much distance you save by serving them on one trip instead of two; merge pairs in order of largest saving, as long as vehicle capacity allows. A University of Southern Denmark course still teaches it as the classic constructive heuristic for capacitated vehicle routing.

It worked where exact methods did not: routes could be computed by hand or with early computers, and the method became a standard building block — a strong starting solution that later metaheuristics like 2-opt, tabu search and genetic algorithms refine.

Merge the pairs that save the most distance first

The savings measure made the greedy choice the right choice

It ran without expensive computing power

Capacity constraints were handled naturally during merging

Its output seeded decades of better algorithms

When the exact answer costs too much, a greedy rule on the right measure ('savings') gets you most of the value — and you can compute it today.

The Clarke-Wright savings algorithm became one of the most cited ideas in vehicle routing and remains a baseline in textbooks and software today, often as the starting point for route optimization in delivery fleets.