Data sources

Fourteen tables in data/logistics/data/:

customers (200)

Customer master: name, contract type (Contract / Dedicated / Spot), credit terms, status, annual revenue potential.

trucks (120)

Tractor master: make, model year, VIN, tank capacity, status, home terminal.

trailers (180)

Trailer master: type (Dry Van / Refrigerated), length, status.

drivers (150)

Driver master: name, hire/term dates, license, CDL class, years experience, home terminal.

routes (58)

Lane master: origin–destination pair, typical distance, base rate, fuel surcharge rate, transit days.

facilities (50)

Pickup/delivery sites: terminals, distribution centers, cross-docks, warehouses with city/state/dock count.

loads (85,410)

Commercial side of each shipment: customer, route, weight, pieces, line haul, fuel surcharge, accessorial.

trips (85,410)

Operational side: driver, truck, trailer, dispatch date, actual miles, gallons used, idle hours. 1:1 with loads.

fuel_purchases (196,443)

Fuel-card transactions: trip, truck, driver, location, gallons, price/gallon, total cost.

maintenance_records (2,921)

Service events: type, odometer, labor + parts cost, downtime hours.

delivery_events (170,821)

Pickup/delivery scheduled vs actual times, detention minutes, on-time flag.

safety_incidents (171)

Incidents: type, fault, injury, vehicle/cargo damage, claim amount, preventability.

driver_monthly_metrics

Pre-aggregated monthly summaries per driver (not currently used).

truck_utilization_metrics

Pre-aggregated monthly summaries per truck (not currently used).

Key formulas

Total revenue

line_haul + fuel_surcharge + accessorial — across loads in scope.

$ / mile

Total revenue ÷ actual miles driven on those loads.

Gross contribution (GC)

Revenue − fuel cost (and − maintenance for trucks). Driver pay, insurance, lease, overhead are not deducted.

GC / mile

Gross contribution ÷ miles. The closest thing to lane / customer / truck profitability we can compute from this data.

Fuel recovery

Fuel surcharge billed ÷ actual fuel cost. <1.0 means surcharge under-collects diesel; we absorb the gap.

OTD (On-Time Delivery)

Share of delivery events arriving at or before scheduled time. Pickup punctuality is reported separately on the Service quality pane.

Detention

Minutes spent at the dock past scheduled time, per event. Reported flat (mean) on entity tabs; total hours on the Service quality pane.

MPG

Miles ÷ engine fuel gallons used (from trips.fuel_gallons_used). 5.5–7.5 mpg is typical for class-8 diesel.

Maintenance $/mile

Total maintenance cost ÷ miles. Above $0.15/mi often indicates an aging asset.

Idle / trip

Idle hours ÷ trip count. Long-haul drivers idle overnight for HVAC; >5 hr/trip is high.

Incidents per million miles

Industry-standard normalization for cross-fleet comparison: incidents × 1,000,000 ÷ miles.

Idle cost estimate

idle_hours × 0.8 gal/hr × avg fuel price. 0.8 gal/hr is the industry assumption for engine idle burn rate.

Share

Entity revenue ÷ total fleet revenue (for current period and active customers / all lanes).

Auto-flags on Customers tab

⚠ At-risk

Revenue falling 3 consecutive months and OTD below 50% and fuel recovery below 0.3.

★ Premium

Stable revenue (no falling 3-mo streak) and OTD above 80% and fuel recovery above 0.7.

⏸ Dormant

Status Active but no loads in the most recent 3 months of data.

Period filter

The dataset covers 2022-01 through 2024-12 (36 months). "Last X months" means the trailing X months from the data end (December 2024). All KPIs and sparklines on a tab respond to the period selector.

Drill-down sub-tables (lanes per customer, customers per lane, drivers per truck, etc.) are computed across the full period and clearly labeled. The Service quality and Safety panes are full-period only — period filtering wasn't built for those.

Known limitations

• Customer revenue is unrealistically uniform. The synthetic data was generated with similar load counts per customer (~430 each), giving a coefficient of variation of 6.8% on revenue. Real freight customers follow a Pareto distribution where the top 10 accounts contribute 50–70% of revenue.
• 1 trip = 1 load. No multi-stop or LTL is modelled. In production, trips would carry multiple loads or split a single load across trucks.
• Costs are partial. Driver pay, insurance, lease, and overhead are not in the data, so "gross contribution" is a contribution toward fixed costs, not net margin.
• Deadhead is not directly tracked. No empty-mile field exists. Lane and truck profitability assumes all miles are loaded miles.
• Detention is mean only. We don't have per-event start/end timestamps for dock occupancy beyond the detention_minutes field.
• Fuel surcharge baseline assumed. Fuel recovery uses the contract fuel_surcharge_rate directly; the true industry comparison would require the DOE/EIA diesel index history.

Build pipeline

Aggregates are pre-computed by Python scripts in scripts/ and saved as JSON in data/logistics/. Re-run any script to refresh the dashboard:

• build_customers.py → customers.json (master + monthly + lanes + facts per customer)
• build_lanes.py → lanes.json (master + monthly + top customers per lane)
• build_trucks.py → trucks.json (master + monthly + top drivers / lanes per truck)
• build_drivers.py → drivers.json (master + monthly + top trucks / lanes per driver)
• build_fuel.py → fuel.json (fleet summary, by-state, by-truck, by-customer-recovery-gap)
• build_summary.py → summary.json (fleet monthly P&L roll-up)
• build_service.py → service.json (OTD + detention by month / facility / customer)
• build_safety.py → safety.json (incidents by month / type / driver / truck + flat list)

The page itself is plain HTML + CSS + vanilla JS, no build step or framework.