Route planning software: turning stops into feasible runs

The model behind the map

An optimiser needs demand as stops with quantities and constraints, resources as vehicles with capacity and cost profiles, a travel-time model, and rules connecting them. Constraints do the real work: opening hours and booked windows, vehicle restrictions by weight, height or emissions zone, driver shift length and break requirements, compatibility rules where a customer or product needs particular equipment, and precedence where collections must follow deliveries. Each omitted constraint transfers a decision from the software to a driver, who will resolve it in whatever way costs least effort at the roadside.

Service time is the input everyone guesses

Travel-time data is usually bought and is reasonably accurate. Time spent at the stop, parking, waiting, unloading, obtaining a signature, handling returns, is normally estimated once and never revisited, yet in multi-drop work it dominates the day. Operations that capture actual arrival and departure timestamps and feed them back into per-customer service times see plan adherence improve without changing the optimiser at all. Where those timestamps do not exist, the honest response is to plan with deliberate slack rather than to trust a tight schedule built on assumptions.

Strategic territory design versus daily routeing

Two different exercises share the same software. Territory or master-route design fixes which customers belong to which day and which vehicle, using historical demand, and is reviewed periodically; daily optimisation solves tomorrow with today's orders. Fixed territories give drivers familiarity, customers consistent times, and planners a simpler day, at the cost of some efficiency. Dynamic routeing extracts more from the fleet but changes arrival patterns constantly, which some customers will not accept. Choosing between them is an operating model decision that the software then serves.

Why plans decay after the gate

Plans fail through late loading, traffic events, refused deliveries, quantity changes and stops added after publication. Software helps in two ways: by producing plans robust enough to absorb small disturbances, and by supporting partial re-planning during the day. It cannot help if the plan was never realistic, and the diagnostic is straightforward. Compare planned against actual sequence, arrival and duration by stop over several weeks; if drivers systematically resequence, the constraint set is wrong, and no amount of optimisation weighting will fix a wrong constraint.

Frequently asked questions

Will optimisation cut mileage on an experienced operation?

Sometimes less than expected on individual runs, because good drivers already route well locally. The larger gains usually come from rebalancing work between vehicles and days, which no individual driver is positioned to do.

How much does map and restriction data matter?

Considerably for larger vehicles. Routes that ignore height, weight, width or access restrictions look efficient and are undriveable, and drivers who divert once stop trusting the plan, which removes the benefit across every other stop.

Should delivery windows be promised to customers?

Only as tightly as the operation can hold. Narrow windows constrain the optimiser sharply and raise vehicle requirements, so windows are a commercial decision with a direct cost that should be quantified before being offered.