Case study · trucking
Building an operations layer for a regional trucking company.
- Client
- A regional trucking company. Named only with written permission, so details that could identify it are left out.
- Status
- In implementation Findings and engineering are real; results are not measured yet.
- Systems
- A desktop TMS with no API, telematics GPS, a fuel-card feed, text messaging
Detention was billed on 0.36% of loads.
110 of 30,652 loads since 2022 carried a detention charge. Trucks wait at docks; the question was whether the business could even see it.
What it cost the business
Unknown, and we do not estimate it for you. The audit found something more basic: the TMS held no arrival or departure time on any of its 135,631 stops, and 80.2% of 2026 loads had a pickup time of exactly 06:00, a default rather than an appointment. With no clock, nobody could bill time they could not prove.
Before
- Detention billed only when someone remembered and could prove it.
- The TMS data stuck on screen: no API, no stop times.
- Fuel bought wherever the driver chose; no per-truck price view.
A data layer first. Then engines, tested.
| Component | What it does | Status |
|---|---|---|
| Data path out of a legacy TMS | Attended exports from a desktop TMS with no API, each one checked against the system’s own on-screen totals. 42,977 trips reconciled. | Built |
| Detention measurement | GPS dwell matched to TMS stops and appointment times, a free-time rule, and a candidate list for dispatch. Engine tests: 35/35. | Built |
| Fuel-card feed | 10,548 fuel transactions joined to trucks for a per-truck price view. | Built |
| Fuel-stop recommender | The cheapest in-network stop on the route, net of the card discount. Passed 30/30 test inputs designed to catch mistakes. Not yet sending anything to drivers. | Testing |
| IFTA miles | Per-state miles from GPS for management review; an estimate to check filings against, not a filing. | Built |
| Fuel anomaly checks | A tank ledger that flags fills with no unit number and refuels that cannot be right, for a person to review. | Built |
| Driver messaging | Text messages only to drivers who agreed to get them, following the phone companies’ texting rules. | Built · not live |
GPS dwell includes legitimate loading time, early arrivals and breaks. That is why the engine produces candidates for dispatch to confirm, and nothing is billed automatically.
The system finds. Dispatch decides.
| The system does | A person does |
|---|---|
| Matches GPS dwell to stops and flags time past free time | Confirms each candidate before any bill is raised |
| Recommends the cheapest in-network fuel stop | Approves the rule before any driver receives a message |
| Estimates miles by state from GPS | Files IFTA from the official records |
| Flags fuel fills that do not add up | Looks at each one; a flag is not an accusation |
What it took from the client
- TMS exports. Load and stop exports from a desktop TMS with no API. Each export was run attended and checked against the system’s own on-screen totals.
- Read-only GPS access, which the client can revoke, and the fuel-card feed.
- A person in dispatch to confirm detention candidates before any bill is raised. A first sample is the next step.
Findings now. Results when they are real.
This engagement is in implementation. The table shows what is measured today and what is not, so you can hold us to it.
| Measure | Before | After / now | Status |
|---|---|---|---|
| Loads with detention billed | 110 of 30,652 since 2022 (0.36%) | Not measured yet | Baseline set |
| Stops with an arrival time in the TMS | None of 135,631 | GPS now provides the clock | Measured |
| Detention candidates past free time | — | 243 candidates out of 1,098 stops, from 71 days of GPS data, for dispatch to confirm | Candidates, not billed |
| Net fuel price against the fleet median | Being measured | Not measured yet | In implementation |
| Hours of manual data entry | — | Not measured yet | Proposed |
What is next
The detention method is written up step by step, so you can check it on your own data. Next, dispatch confirms a sample of detention candidates so the true billable share is known. Fuel recommendations go to drivers who opted in, and net fuel price against the fleet median is measured month by month. Both results will be published here with their method, whatever they show.
Published October 2026. Findings from a TMS export audit and GPS analysis, September 2026.
Find out what your TMS and GPS data already know.
A one-hour working session on Zoom with the founder. We read your public FMCSA record first; $100 at booking.