Blog · Algorithms
By Adnan Alotaiby · Published 2026-03-26 · Updated 2026-04-11 · 8 min read
Every time you tap "Optimize" in a route planning app, a computer solves one of the most studied problems in mathematics. The Travelling Salesman Problem (TSP) asks: given a list of locations, what is the shortest route that visits each one exactly once? For delivery drivers, this is the core problem behind every route optimization tool on the market.
This guide explains how TSP algorithms work, why exact solutions are computationally impossible beyond a small number of stops, and how modern heuristics produce near-optimal delivery routes in under two seconds. Brute force exhaustive search is impossible — for 50 stops, there are more possible routes than atoms in the observable universe. Instead, SortDrops uses multi-start nearest-neighbor heuristic combined with 2-opt improvement, running on a real road distance matrix from OSRM.
The combination of these algorithms produces routes within 3–8% of optimal, solving 50 stops in under 2 seconds. For fleet dispatch, SortDrops adds k-means geographic clustering to divide stops across multiple drivers before running per-driver TSP optimization.