Network Routing Optimization
Routing decisions moved from optimizing each lane in isolation to optimizing total network cost, over a six-month migration that touched every lane in a multi-region distribution network.
The challenge
A multi-region distribution network optimized every lane on its own, which looked efficient lane by lane and wasteful in aggregate:
- Each lane's routing was minimized for its own cost, with no visibility into how that choice affected other lanes
- Backhauls were frequently run empty because the outbound and return legs were optimized by different teams or systems
- Cross-hauling occurred where two regions independently routed shipments past each other in opposite directions
- Carrier capacity was booked lane by lane, missing opportunities to consolidate volume across adjacent lanes
- No one had a single view of what the network was actually costing as a whole — only what each lane cost in isolation
How it works
Optimizing the network, not the lane
The migration replaced lane-level cost minimization with a network-flow model that considers every lane simultaneously:
- 01
All lane, volume, and carrier-rate data across the multi-region network was consolidated into one routing dataset
- 02
A network-flow optimization model was built to minimize total network cost, not the cost of any single lane
- 03
Backhaul and cross-haul opportunities are identified automatically as byproducts of the network-level solve, not manual audits
- 04
The model re-solves daily against updated volume and rate data, re-routing where total cost has shifted
- 05
Regional teams were moved from independently booking lanes to reviewing network-level routing proposals
- 06
A six-month phased cutover let each region validate the new routing against its own freight budget before full adoption
What we built
Key capabilities
Total network cost, not lane cost
Every routing decision is evaluated against its effect on the whole network, not just the lane it sits on.
Backhauls found automatically
Empty-leg and cross-haul opportunities surface as a byproduct of the daily network solve, not a manual audit.
Daily re-optimization
Routing re-solves against fresh volume and rate data every day instead of being locked to a static plan.
Regional teams stay in the loop
Regions review and approve network-level proposals rather than losing visibility into their own freight.
Before vs after
Routing, before and after
- Optimization unit
- Individual lane → Total network
- Empty / deadhead miles
- Baseline → 19% fewer
- Total freight spend
- Baseline → 15% lower
- Routing cadence
- Static plan → Daily re-solve
Business impact
What it changed
15% lower freight spend
Total network freight cost fell 15% comparing the six months before migration to the six months after full cutover.
19% fewer empty miles
Backhaul and cross-haul opportunities identified by the network-level solve cut deadhead mileage by 19%.
A single routing view
Regional teams now review network-level proposals instead of booking lanes with no visibility into the rest of the network.
Technology stack
“A lane can look optimal and still be wrong for the network around it. Six months in, routing decisions are made against the number that actually matters — total network cost.”
Keep reading