What captured billing is worth, in your numbers.
Every WMS pitch claims it pays for itself. This page gives you the arithmetic instead of the claim, and a way to measure your own leak rate in one week rather than trusting ours.
The three places a 3PL loses money on its own work
- 1Work done and never billed. A pallet broken down by hand, an inspection, a rush, an extra receiving line. It happened on the floor; it did not reach the invoice, because the invoice was reconstructed at month-end from memory and spreadsheets.
- 2Work billed late or disputed. A charge that cannot be traced to a specific event on a specific day is a charge a client can argue with — and the concession usually costs more than the fee.
- 3Labor spent assembling invoices. Days per month of somebody senior chasing numbers, rather than minutes reading a report that was already true.
Amphora addresses the first two the same way: the fee is recorded as a billable event at the moment the work happens, priced by that customer's rate card, and every event points at the movement that caused it. The third follows from the first two.
Measure your leak rate — one week, no commitment
Do not take a number from a vendor's brochure, including this one. Run a pilot week: receive and ship a slice of real work through Amphora alongside your current process. At the end of the week, compare the billable events Amphora captured against what your invoicing process captured for the same work. The difference is YOUR leak rate, on YOUR book, and it is the only number worth building a business case on.
Published leakage numbers for 3PL billing are scarce, inconsistent, and usually produced by someone selling something. We would rather hand you the instrument than a statistic you would be right to distrust.
The arithmetic
Fill in the middle column from your own book. The right column is one worked example — a mid-size operation — shown so the shape of the calculation is unambiguous.
| Line | Yours | Worked example |
|---|---|---|
| A. Monthly revenue billed to clients | $180,000 | |
| B. Measured leak rate (from your pilot week) | 3% | |
| C. Revenue recovered per month (A × B) | $5,400 | |
| D. Hours per month assembling invoices | 24 hours | |
| E. Fully-loaded cost of those hours | $45/hour | |
| F. Labor recovered per month (D × E) | $1,080 | |
| G. Disputed charges written off per month | $900 | |
| H. Share of those disputes that were unprovable, not wrong | 60% | |
| I. Disputes recovered per month (G × H) | $540 | |
| Monthly value (C + F + I) | $7,020 | |
| Annual value | $84,240 |
The worked example uses a 3% leak rate purely to show the arithmetic. If your pilot week measures 0.5%, use 0.5% — the calculation still works, and the answer is still yours rather than ours.
What it costs on the other side of the ledger
- Software. Self-hosted, no per-seat meter, no per-transaction toll. Pricing is a conversation, not a matrix — ask, and you will get a number rather than a form.
- Hardware. One small server. The benchmark in the capability sheet runs a warehouse of two million ledger movements on an eight-core machine; you likely have one spare.
- Your time. An onboarding afternoon per client, plus somebody who owns the server: patches, backups, and a plan for the day a disk fails.
- Switching cost. Real, and worth naming: retraining the floor, running parallel for a period, and moving master data. The pilot exists so you find out what that costs before you commit to it.
The second-order value, which is harder to price
A count you can defend ends an argument with a client in one screen instead of an afternoon. A recall becomes a lot filter rather than a war room. An SLA the system enforces is a promise your account manager stops apologizing for. None of these belong in the table above — but they are usually what the operations lead remembers six months in.
Bring one month of your billing and one client's rate card to a call, and we will build this worksheet with your real figures on the screen. If the number is small, we will say so — a pilot that fails cheaply is worth more to you than a rollout that fails expensively.
parth@brandboxai.app