You Can't Collect What You Can't Find

Invoice 8842 goes out for $41,600 of hauling. The check arrives for $38,700, with a remittance page listing eleven load numbers and one sentence: unable to verify delivery, please provide signed tickets.

Nobody disputes a rate. Nobody claims the loads didn't run. The customer's AP clerk is doing what her procedure tells her to do, which is pay what she can tie to a signed receiving document and hold the rest. The eleven loads she's holding are worth about $2,900. Whether you collect that money has almost nothing to do with whether the work happened, and almost everything to do with how your office filed paper four months ago.

Your filing answers a question nobody asks

Walk the request through your own building. Eleven serial numbers, spread across a job that ran from late April into June.

The scale copies are in the scale house, on a roll, in printing order. The driver copies came in as folded batches and got filed by the week they arrived. The customer-signed copies are the ones you need, and those are the ones that traveled: signed at the gate, ridden back in a cab, turned in on a Thursday with sixty other tickets from four other jobs.

So the request is addressed one way and your paper is organized another. AP asks by serial, customer, job, and date range. Your file is sorted by the day it landed on a desk. That mismatch is the whole problem, and it isn't a discipline failure. The clerk filed by turn-in date because that's the order the paper arrives in, and no filing key is obvious at the moment a ticket comes through the door.

Six hours later she has eight of the eleven. Two more surface in a different month's folder, right serial, no customer signature on either. One never turns up at all.

What the eight cost and what the three cost

The clerk time is the part that shows up somewhere. Six hours of a billing clerk at $28 an hour is $168, plus the invoices she didn't build that day.

The real number is the three loads. You copy the eight, credit the three, and $790 leaves as a credit memo that books as a billing adjustment. On the books that looks like housekeeping. Those loads were hauled, dumped, and signed for by a foreman who is now working somewhere else.

Here's the claim your controller will want to argue with: the eight you produced cost you more than the three you conceded, and the bill for them comes later.

The customer's AP department now has a data point. One short-pay against your invoices returned a credit, on the first attempt, with no escalation and no argument. Nothing about that requires anyone over there to act in bad faith. It requires a procedure that says hold unverified lines, and a supplier who has just demonstrated that some share of held lines never gets defended. The next invoice gets the same treatment, and so does the one after that, because the practice paid. Your clerk will spend another day in the folders next quarter, on a longer list.

flowchart TD
  A["Short-pay: 11 loads held on invoice 8842"] --> B{"How is the paper filed?"}
  B -- "by turn-in date" --> C["Six hours of reading, 8 of 11 defensible"]
  C --> D["Credit 3 loads, $790 conceded"]
  D --> E["AP procedure learns the short-pay works"]
  E --> A
  B -- "by every handle at once" --> F["All 11 copies produced the same day"]
  F --> G["Every held line answered from the file"]

The handles have to ride with the ticket

Paper filing can't solve this, and a stricter rule about turn-ins won't either. Any single filing key fails, because a dispute can arrive addressed by any of four handles. A stack sorted by serial has nothing to say about one customer's April. A stack sorted by customer has nothing to say about serial 40912.

A ticket captured at turn-in carries its own handles: the serial it was printed with, the date hauled, the customer and job it belongs to, whether a signature is on it, and eventually the invoice line it became. Digitize tickets as crews and drivers hand them in, and April's eleven loads are already tied to the job that ran them and the invoice that billed them. Nobody has to reconstruct that in September.

The bigger effect lands in April, though, not in September. Today you learn a copy is unsigned when AP asks for it, and by then the foreman who took the load has changed employers. A captured ticket with no signature on it is a non-compliant ticket that same week, while the driver still remembers which gate he pulled into and a phone call still fixes it. Some of those eleven were probably never signed at all, and nobody in your office knew.

Price is the other half of a defense. When the rate lives in a billing rule with an effective date, the answer to "why is this load $264" is a rule you can point at, not a clerk reconstructing what the contract said in April. Build invoice totals only from tickets that priced against a rule, and you can walk any line of them back to its rule months later without a search. That's the narrow thing Termco does: capture the tickets, flag the ones that don't reconcile, hold billing rules and reference data where billing can see them, and build totals from nothing else.

A measurement is already sitting in your own files, and it costs about an hour. Ten load numbers off an invoice from four months ago, handed to whoever answers AP calls, against a timer. Two numbers come back: how many signed copies she found, and how long it took her to find them. That's your collection rate on any line a customer decides to hold. Nothing on your aging report shows it, which is why the credit memos keep looking like housekeeping.

If you'd rather see what a fresh reader finds in your paper, send 10-20 sample tickets to hi@termco.ai and we'll send back what we find within 48 hours.