Logistics · Intelligent Automation
Automating carrier rate selection for 2,200 monthly shipments — cutting cost per booking by 31%
A mid-size US freight broker managing 2,200+ shipments per month across parcel and LTL carriers was spending 40% of its operations team's time manually pulling rates from carrier portals and booking shipments. The process was slow, inconsistent, and leaving margin on the table. We built an automated rate intelligence engine on top of their existing TMS that selects the optimal carrier per shipment in under 4 seconds, reduced cost per booking by 31%, and freed 18 hours of ops capacity per day — without replacing a single system.
Business Context
The margin wasn't being lost to the market.
It was being lost to the process.
The broker had strong carrier relationships and competitive contract rates across 11 carriers — a mix of national parcel, regional parcel, and LTL providers. On paper, the rate structure was solid. In practice, the ops team was not consistently accessing the best rate for each shipment. With 2,200 shipments per month and a 50-person operations team, rate shopping was done manually: open the carrier portal, enter shipment details, note the rate, repeat across 4–6 carriers, then book. The process took 12–18 minutes per shipment on average. At that volume, rate shopping was consuming the equivalent of 4.5 full-time employees every single day.
What the manual process was costing
- 14 min
- average time per rate shop
- 40%
- of ops capacity consumed
- $6.20
- avg overspend per shipment
Across 4–6 carrier portals per shipment, manually entered
Rate shopping and booking across a 50-person team
vs. optimal rate — driven by inconsistent carrier selection
The inconsistency was the deeper problem. Different ops staff applied different carrier preferences — some defaulted to familiar carriers regardless of rate, others prioritised speed over cost even when the client SLA did not require it. There were no enforced business rules. A shipment that should have gone LTL was going parcel. A time-sensitive delivery was being booked on a slower service because the rate looked lower on the surface without accounting for accessorial charges.
The TMS they were running — a mid-market platform with carrier integrations — had a rate shopping module, but it was limited to 3 of their 11 carriers and required manual triggering. It was not being used consistently. The gap between what the system could theoretically do and what was actually happening on the floor was where the margin was disappearing.
Scope of Work
What we were asked to build
Carrier API integration layer
Direct API integrations with all 11 carriers — parcel, regional parcel, and LTL — normalising rate responses into a unified schema regardless of carrier format. Real-time rate retrieval with sub-2-second response targets per carrier.
Rate intelligence and selection engine
Rules-based selection engine applying client SLA tiers, shipment characteristics (weight, dimensions, origin/destination, hazmat flags), and carrier performance history to rank and select the optimal carrier automatically. Configurable business rules per client account.
TMS integration and auto-booking
Bidirectional integration with their existing TMS — pulling shipment data on creation, pushing the selected rate and carrier back, and triggering booking automatically for shipments below a configurable value threshold. Human review queue for exceptions.
Rate analytics and savings dashboard
Operations dashboard surfacing carrier performance, rate variance, savings per shipment, and exception patterns — giving ops managers visibility into where the engine is performing and where manual overrides are occurring.
Constraints we worked within
- TMS could not be replaced or significantly modified — integration had to be additive via API
- Carrier API credentials and rate agreements were commercially sensitive — required strict access controls
- Auto-booking threshold required sign-off from each client account — rollout was phased by account
- LTL carriers had inconsistent API maturity — two required EDI integration rather than REST
Explicitly not in scope
- Last-mile or final-mile carrier integrations
- International freight or customs documentation
- Client-facing shipment tracking portal
- Carrier contract renegotiation or procurement advisory
System Architecture
One rate engine. Eleven carriers. Zero manual bookings for 84% of shipments.
How We Worked
6 months. 3 phases. One number that mattered: cost per booking.
Integration & Rules Architecture
Carrier API discovery and integration for 9 of 11 carriers. EDI setup for the two legacy LTL carriers took longer than anticipated — completed in month 2. Business rules workshops with ops managers to codify carrier selection logic that was previously in people's heads.
Engine Build & TMS Integration
Rate intelligence engine built and tested against 6 months of historical shipment data. TMS integration completed. Auto-booking logic built with configurable thresholds. Exception queue UI built for ops team review.
Controlled Pilot
Launched on 3 client accounts representing 400 shipments/month. Ops team ran parallel manual checks for the first two weeks. Auto-booking accuracy validated at 96.4% against what the ops team would have selected manually.
Full Rollout & Optimisation
Expanded to all client accounts. Ops team transitioned from booking to exception management. Savings dashboard live. Carrier performance data feeding back into selection rules on a weekly refresh cycle.
Working rhythm
- CadenceTwo-week sprints, weekly ops team syncs
- Decision ownerVP of Operations and Head of Technology
- Primary metricCost per booking vs. pre-automation baseline
- Escalation SLA24 hours with written recommendation
Results
Measured at 90 days post full rollout.
reduction in average cost per shipment booking
Was: $6.20 average overspend per shipment vs. optimal rate
Driven by consistent application of business rules across all 11 carriers and elimination of carrier preference bias. LTL optimisation alone accounted for 44% of total savings — shipments previously defaulting to parcel were correctly routed to LTL where the rate and SLA both supported it.
of daily ops capacity recovered across the team
Was: 40% of a 50-person ops team consumed by manual rate shopping and booking
The same team now manages exceptions, client escalations, and carrier relationship management — higher-value work that was previously crowded out by manual booking volume. No headcount reduction; capacity was redeployed.
average time from shipment creation to carrier selection and booking
Was: 12–18 minutes per shipment across manual portal checks
For shipments below the auto-booking threshold — approximately 84% of total volume — the entire rate shop and booking cycle is now fully automated. The remaining 16% route to the exception queue for human review, typically resolved within 8 minutes.
auto-booking accuracy vs. what ops team would have selected manually
Was: no consistent selection standard — carrier choice varied by individual
Validated during the 2-week parallel run in the pilot phase. The 4% of cases where the engine and ops team diverged were reviewed — in the majority, the engine's selection was defensible on cost grounds; a small subset informed rule refinements before full rollout.
What This Means for You
The rate intelligence engine we built here is not unique to this broker. It applies to any logistics operation where carrier selection is still a manual decision made by individuals applying inconsistent rules.
- 01Your ops team spends more time booking shipments than managing exceptions and client relationships
- 02Carrier selection varies by individual — there is no enforced business logic applied consistently across all bookings
- 03You have contract rates with multiple carriers but no system ensuring the right carrier is selected for each shipment
This engagement was scoped as an additive automation layer on top of an existing TMS — no platform replacement, no disruption to live operations, no carrier relationship changes. Six months from kickoff to full rollout across all client accounts.
See how we approach Intelligent Automation for logistics