How to calculate coordination time per shipment, and check the math
Key Takeaways
- Coordination time per shipment is the sum of the minutes spent on each task a load needs: quote, booking, pickup confirmation, follow-up, tracking updates and POD.
- Multiply it by a real shipment count from the same period, not a forecast.
- The per-shipment figure must reconcile with total hours the team actually spends; if it doesn't, one of the two numbers is wrong.
- A per-shipment minute figure is an assumption until you time it on your own loads.
- Measure the same lane or process before and after automation, once the workflow is running on real loads.
The formula, in one line
Coordination time per shipment is the minutes a person spends on every task one load requires, added together; time saved is that figure multiplied by the number of shipments the automation actually handled. The calculation is simple. What makes it credible is checking it against total hours and real shipment counts.
Coordination time here means the human work around a load, not transit time. It covers asking for quotes, comparing them, tendering (formally offering the load to a carrier), getting the rate confirmation, confirming the pickup, chasing ETAs, answering status questions and collecting the POD (proof of delivery). It excludes warehouse work like picking and loading.
Which tasks belong in the count?
Count every touch a coordinator makes on a load from the moment the order is released to the moment the POD is filed. Leaving out small tasks is the most common reason estimates come out too low.
Start with a task list and time each one on a sample of real loads:
| Task | What to time | Easy to miss |
|---|---|---|
| Order intake | Reading the order, creating the shipment | Fixing missing weights or addresses |
| Quote | Requesting and comparing carrier quotes | Follow-ups to carriers who didn't answer |
| Booking and tender | Sending the tender, getting the rate confirmation | Getting approval above a cost limit |
| Pickup confirmation | Agreeing and reconfirming the window | Rescheduling when the time doesn't work |
| Tracking and updates | Check calls, ETA emails, internal updates | Customer "where is it?" emails |
| POD | Requesting, filing and chasing it | Second and third requests |
Time loads that went smoothly and loads that didn't. An average built only from clean loads understates the work, because exceptions take far longer than routine steps.
A worked example with hypothetical numbers
The example below is hypothetical, with round numbers chosen to show the method. It does not describe any Cartage customer.
Suppose a distributor times 20 loads and finds this average per shipment:
- intake: 2 minutes
- quote: 4 minutes
- booking and tender: 2 minutes
- pickup confirmation: 1 minute
- tracking and updates: 2 minutes
- POD: 1 minute
That's 12 minutes per shipment. If the distributor ships 600 loads in a quarter, coordination time is 600 × 12 = 7,200 minutes, or 120 hours for the quarter.
If automation handles 450 of those 600 loads and the team still spends 3 minutes on each of them (approvals, exceptions), the time saved is 450 × (12 − 3) = 4,050 minutes, or 67.5 hours. The other 150 loads still cost the full 12 minutes each.
Notice what the example needs: a timed per-task figure, a real count of loads handled, and the residual time people still spend. Leave out any of the three and the result is a guess.
How do you know the per-shipment number is right?
Run three sanity checks: reconcile with total hours, compare against headcount, and test the number on loads that went wrong. A per-shipment figure that fails any of them needs to be re-timed.
1. Does it reconcile with total hours?
Multiply your per-shipment minutes by the shipment count for a period, then compare with the hours the team actually spent on freight in that period. In the hypothetical above, 120 hours a quarter is roughly 9 hours a week. If two coordinators say freight fills most of their week, 12 minutes per load is far too low, or the shipment count is missing loads.
The two numbers will never match exactly, but they shouldn’t be off by a multiple. A big gap usually means either missing tasks in the per-shipment list, or freight work hidden inside other roles such as customer service or purchasing.
2. Does it survive exceptions?
Take five loads that had a problem: a declined tender, a missed pickup, a late delivery, a missing POD. Time them separately. If the average for those loads is several times the clean-load figure, weight your average by how often exceptions actually happen.
3. Is the count real?
Use shipments that actually moved in the period, from the ERP or the shipment record, not orders booked or planned. A count that includes cancelled orders or double-counts split shipments inflates any total.
For which metrics to track alongside time, see freight coordination KPIs.
How to treat a published per-shipment assumption
Treat any published minutes-per-shipment figure, including ours, as an assumption to test against your own timing. Use it to frame the conversation, then replace it with your measured number.
A plastics and rubber manufacturer is a useful case. Wilson, by Cartage, an AI logistics agent, has scheduled 2,800+ shipments for that company, processing its daily SAP batch files, routing carrier emails and assigning pickup times. When we model coordination time for work like that, one assumption we use as a starting point is 5 minutes of coordination time per shipment.
That 5-minute figure is an assumption, not a measured result, and we don't publish a total-hours figure for this customer. To test it on your operation, time the same kind of work (reading the order, getting carrier replies onto the right load and agreeing a pickup time) on 20 of your own loads. Whether your average comes out higher or lower, replace the assumption with your measured number.
Also check what the shipment count actually covers. 2,800+ is a count of shipments scheduled. Before you multiply any minute figure by a count, make sure both describe the same tasks, or the total will mix work that was handled with work that wasn't.
When should measurement start?
Start measuring once the workflow is running on real loads, and compare the same lane or process before and after. Measuring during setup mixes onboarding effort with steady-state work.
Wilson’s own guidance is to compare the lanes it runs with how those lanes ran before, using manual touches per load, coordinator hours, time to book, missed pickups, time from exception to notification and POD collection time. Coordinator hours are only one of those. A lane where hours fall but missed pickups rise hasn’t improved.
Every message, confirmation and change Wilson makes on a shipment is recorded in its change history. That gives you a load-by-load record of what was done and what was escalated, which makes the “residual minutes per load” in the formula easier to check. For building the case around those figures, see a freight automation business case per load.
The check that decides it
A per-shipment number is credible when it's timed on your loads, multiplied by a real count and close to the hours your team actually reports. If you've timed 20 loads already, bring that sheet and we'll rebuild the per-shipment math with you.
Sources
- Cartage customer results, October 2026 (anonymized)
- Cartage, Wilson product FAQ, October 2026: https://cartage.ai