In commerce and logistics, the Transportation Management System is where margins are quietly won or lost. It sits between orders and carriers and makes thousands of small decisions a day — which carrier, which service level, which route — and at volume, the quality of those decisions is the difference between a healthy shipping line and a leaking one. I've spent a good amount of my career building in this space, and the same few levers come up every time.
Routing optimization
The most visible win is routing: consolidating shipments, sequencing stops, and choosing lanes that balance cost against delivery commitments. The math is constrained — service-level promises, cutoff times, capacity — and a good TMS turns what used to be a dispatcher's judgment call into a repeatable, optimized decision.
Rate shopping and carrier selection
For every shipment there's a cheapest acceptable way to move it, and finding it by hand doesn't scale. Rate shopping compares carriers and service levels in real time against negotiated rates; carrier selection then applies the business rules — preferred carriers, zone coverage, performance history — to pick the winner. Done well, this is where the "shipping cost savings" everyone wants actually come from, shipment by shipment.
Logistics rules and automation
Most of a logistics operation's administrative load is rules that rarely change: how to handle a region, a product class, a customer tier, an exception. Encoding those rules into the TMS removes the manual handling and, just as importantly, makes the behavior consistent and auditable instead of living in someone's head.
The architecture that holds at scale
A TMS that's correct but slow is still a bottleneck. At real volume — millions of shipments — the system has to make these decisions reliably under load, which is an event-driven, service-oriented problem: rate requests, selection, and label generation as discrete steps that scale independently and fail gracefully. This is the part that separates a working prototype from a system a business can actually run on, and it's the same production discipline — evals on decision quality, observability, rollback — that any system touching real money and real inventory demands.
Where the savings really come from
The headline is usually cost savings, but those are downstream of something simpler: replacing thousands of manual, inconsistent decisions with fast, optimized, rule-driven ones. Get the routing, rate shopping, and rules right on an architecture that holds at scale, and the savings — and the better customer experience — follow.