Put It on the GC's Account

"Put it on Carver's account. It's their job."

The driver saying it hauls for Miller Excavating, the dirt sub on the Lincoln Street school. The operator in your scale house has a line of trucks and no reason to doubt him. Carver Construction is the GC, Carver's name is on the fence, and Carver has carried an account with you since 2011. He writes Carver in the customer line, the driver signs, and the truck is rolling in under a minute. Over the next two weeks the same sentence loads 23 trucks: 540 tons of 57s and base rock, $7,425 at Carver's negotiated $13.75.

The invoice goes out with the month's billing. Two weeks later it comes back. Not short-paid, not argued, refused, with one sentence from Carver's AP office: per the subcontract, site aggregate is furnished by the subcontractor. Carver never ordered the stone. Miller did, through a foreman on a cell phone, and Miller's subcontract says the stone is Miller's cost.

So now you bill Miller, forty days after the first load. Miller has no account, so a credit application goes out before an invoice can. Miller's PM counts 19 loads, not 23, because four tickets were signed by a foreman who has since moved to another job. And Miller never agreed to $13.75, because $13.75 is Carver's price, earned by Carver's volume. Your book has Miller at list, $15.25, which Miller's PM reads as a penalty for getting caught. By the time a check lands you're 90 days past the first load, and the stone left your pit exactly as ordered, every ton of it.

One name, three decisions

The customer line looks like a label. It's an assignment of debt. The weight says how much left your property, the product says what it was, and the name says who owes for it, and the name is the only one of the three the scale can't measure.

Everything commercial rides on it. The name picks the invoice address, and with it the price, because rates hang off the account, and the terms, because net 30 hangs off the account too. Three decisions, made in four seconds at a window, by a driver who didn't order the material and an operator who has never seen the subcontract. Neither of them is careless. They're just not party to the information, and the window has no way to check. Your scale is certified to fifty pounds; the customer field runs on hearsay.

Then it hardens. Once the name is keyed, no downstream step ever re-asks it. The invoice run doesn't ask whether Carver owes. It asks what Carver's price is. A wrong weight at least has a chance of looking wrong to a clerk who knows what a tandem carries. A wrong name looks perfect, because Carver really is on that job, buying other stone, on other tickets, at the same gate.

And the driver's signature proves the load was received, not who owes for it. Receipt happened at your scale at 9:40 on a Tuesday. The debt was settled weeks earlier, in a buyout meeting you weren't invited to, and wrote itself into a subcontract nobody at the window will ever read.

A job is one name wearing four companies

This isn't a rare mix-up, because a job site is one name wearing several companies. Owner, GC, a dirt sub, a utility sub, a trucking broker, an owner-operator painted in the sub's colors. Orders come in by phone from foremen who name the job, never the entity: send four loads of bedding sand to the school. Bedding sand is in Miller's scope. Base rock under the parking lot is in Carver's. Same gate, same pile, different debtor by product, and the scope split lives in a subcontract exhibit, not at your scale.

So the window does what any sane window does: it defaults. The first company to open credit on a job becomes the job's name, and every later order inherits it. That default is invisible all season, right up until an invoice tests it.

No load moves on a job name

Here's the claim your dispatcher will fight: a job name is not an order. No truck gets loaded until the order names the company that owes, and the window keeps a card for every active multi-company job saying who may charge what. Lincoln Street school: Carver pays 57s and base, Miller pays bedding sand, anyone else is cash.

The objection is honest. Orders arrive on the radio, foremen don't talk in legal entities, and the line is forty trucks deep; nobody wants the scale operator cross-examining drivers about buyout scopes. But the card isn't for interrogation. It moves the question from the window, which can't answer it, to the order desk, which can, with one phone call per order instead of one dispute per month. The operator just reads the card, which is faster than guessing.

And watch the default some owners actually run, said out loud or not: when in doubt, bill the GC, they have the money. It's exactly backwards. The biggest company on the job has the most formal AP and refuses fastest, and every refusal restarts your terms clock at the worst possible distance from the work.

flowchart TD
  A["Foreman orders by job name"] --> B{"What does the window write?"}
  B -- "whatever the driver says" --> C["23 loads keyed to Carver"]
  C --> D["Invoice refused: material by sub"]
  D --> E["Rebill Miller, 40 days late"]
  E --> F["Rate and load count disputed"]
  B -- "job card: payer by product" --> G["Keyed to Miller from load one"]
  G --> H["Invoice ages from day 30"]

The card's weakness is the same as every taped-up list: it's current until the scope changes, and scopes change on the radio. That upkeep is the slice Termco takes on. Which companies buy on which jobs, and which products each one pays for, lives in reference data instead of on the glass. Tickets get digitized the day they're written, exactly as written. A billing rule checks each name against the job's buyer list, so a load keyed to the wrong company gets flagged within days, while the foreman who ordered it still answers that phone. Invoice totals then assemble per payer. Nothing changes at the window: same book, same pen, same four-second answer through the glass. If you want to see what that check finds in your own paper, send 10-20 sample tickets to hi@termco.ai and findings come back within 48 hours.

Pull the rejections

The test costs one evening. Pull a year of invoices that came back refused or credited with some version of not ours, and price the time between first load and final payment on each. Then take your busiest multi-company job this season and list every entity you billed on it, next to what the subcontract scopes actually say. Anywhere one company's name covered another company's debt, you found this leak, and odds are the window has been covering for it with a default nobody chose.

The stone and the weight were right, and the signature was real. The only fiction on those 23 tickets was one proper noun, offered in good faith and accepted because the line was long. Your operation measures everything that crosses the scale except the one word that decides who pays for it.