Freight is one of the few cost lines where a purely computational improvement produces immediate savings with no change to what you make, who you buy from, or what you promise customers. It’s also one of the least examined, because it tends to sit with whoever manages the warehouse rather than with anyone whose job is cost.
What Routing Decisions Actually Involve
A routing problem isn’t just finding the shortest path. It’s deciding which orders go on which vehicle, in what sequence, leaving at what time, with which driver, subject to vehicle capacity, delivery windows, driver hours, load compatibility, and the reality that a truck which is full by volume may be half empty by weight.
With a handful of drops this is solvable on a whiteboard. Beyond about a dozen, the number of feasible combinations grows past what anyone can evaluate, so planners settle on a route that works and reuse it. Fixed routes are stable and easy to manage, and they are almost always leaving money on the table — because the order profile that justified them has since changed.
Where the Savings Some From
Consolidation. Two part-loads going to adjacent destinations on different days is a common pattern that nobody spots because the orders were entered by different people at different times. Systematic consolidation is usually the single largest source of freight savings available.
Sequencing. Delivery order affects total distance substantially, and the intuitive sequence — nearest first — is frequently not the efficient one once time windows are involved.
Mode selection. Some shipments move by express freight out of habit rather than need. Comparing the true cost of the faster mode against the actual requirement date, per shipment rather than by blanket policy, tends to move a meaningful share of volume to cheaper modes with no service impact.
Backhaul utilisation. Vehicles returning empty are pure cost. Matching return legs against inbound material collections is straightforward to model and rarely done manually because the two flows are managed by different teams.
Keeping Service Intact
The fear with optimisation is that cost savings arrive at the customer’s expense. This is a real risk if the model is given cost as its only objective. It’s avoidable by treating delivery windows and service commitments as hard constraints rather than weighted factors — the optimiser then finds the cheapest solution that still meets every promise, which is a different and much safer question.
It’s also worth building in some slack. A route optimised to the minute leaves no room for traffic, a slow unload, or a driver break running long. Plans that are theoretically optimal and practically fragile get abandoned by drivers within a fortnight.
Starting Small
Take one region and one month of shipments. Run the optimiser retrospectively against what you actually did and compare. You’ll get a defensible number for the opportunity before committing to anything, and you’ll find out whether your address and time-window data is clean enough to rely on.
ticktick.ai optimises routing against live orders and constraints, including inbound collections for backhaul matching.
