Ninety Minutes of Free Rock

Somewhere in your scale house there's a pad of blank tickets for the day the system goes down. Maybe it's a carbon-copy book from an office supply store, maybe it's a stack of photocopied forms in a drawer under the radio. Nobody has looked at it since the last outage. It is the entire disaster recovery plan for the most important machine in your business, and the day it comes out is the day your billing runs on handwriting and trust.

The outage will pick its own morning. A thermal printer dies mid-roll, or the scale software hangs on an update it started overnight. Either way the picture is the same: trucks stacking up between the gate and the scale, a concrete plant on the phone asking where its rock is, and a scale operator holding a pen.

Nobody closes the gate. Holding paying trucks at the fence feels worse than any paperwork problem, so the pen wins.

Outage leakage is the whole rate

Your normal leakage is a percentage. A miskeyed job number here, a product substitution there, and the damage runs a few points against tickets that mostly made it to an invoice. An outage is different in kind: for however long the system is down, the ticket printer that anchors "one ticket, one billable event" doesn't exist. Every load that leaves is billed only if a human reconstructs it later, and reconstruction is a process nobody owns.

Put rough numbers on it. A mid-size pit moving 22 loads an hour at an average ticket of $290 pushes about $6,400 an hour across the scale. A ninety-minute outage puts 33 loads through an improvised paper process. If three of them never make it back into the system, that's roughly $870 gone, and it's gone in the quietest way possible: no voided ticket, no dispute. There's no record that the loads existed, so there's nothing to notice missing. Three or four outages a year and you're a few thousand dollars light with a clean-looking ledger.

The direction of the error is not symmetric, either. A customer who gets billed for a load that didn't happen calls you. A customer who hauled three loads during your outage and got billed for one does not.

Why the pad doesn't save you

Handwritten outage tickets fail in ways printed tickets can't.

They're incomplete. The operator writing by hand while eleven trucks idle at the window captures the truck and the product and skips the tare, the job number, the PO. Every skipped field is a decision somebody in the office has to make later, from memory, about a load they never saw.

They're unsequenced. Your system tickets carry serial numbers, so a gap announces itself. A handwritten ticket that blows off a dashboard or rides home in a driver's jacket leaves no gap anywhere, because no numbering ever said it was written.

And they get keyed late, which is where the duplicates breed. The pad gets keyed Thursday. The scale operator, being conscientious, also re-enters a few loads from memory once the system comes back Tuesday afternoon. Now the same load exists twice under slightly different weights, and whether your month ends short or long depends on which copy the billing clerk trusts. One morning of outage produces every species of bad ticket at once: missing, duplicate, and miskeyed.

Holding trucks, handwriting tickets: those are two of the gate's three options. The third is the worst: waving the regulars through with a promise to catch it up later, from memory or a loader count. Those loads start their life as free rock and have to be argued back into existence.

flowchart TD
    A["Scale system goes down"] --> B{"What does the gate do?"}
    B --> C["Hold trucks until it's back"]
    B --> D["Handwrite tickets on a pad"]
    B --> E["Wave known customers through"]
    C --> F["Every load prints a real ticket"]
    D --> G["Keyed days later, fields missing, some duplicated"]
    E --> H["No record at all"]

The loads left witnesses

If ticketing is down and you can't run a paper process you'd trust with cash, hold the gate. Twenty minutes of idling trucks is a visible, bounded cost. Plenty of operators will call that insane: you never stop a customer's trucks, and the pour won't wait. Fine. Then accept the second-best position: every load that leaves during an outage must leave a second record, made the same day, by someone other than the person keying the pad.

The second records already exist; you just don't collect them. The loadout camera has timestamps. Dispatch knows which trucks it sent and how many turns they ran. The plant totalizer kept counting tons while the office was dark. The customer's foreman counted loads at the other end, because he always does. Any one of these, set against the pad, tells you whether 33 loads left or 36 did. None of them gets looked at, because the person keying the pad on Thursday was never asked to prove the count, only to type what's on the paper.

That's the real finding about outages. The system going down starts the clock, but the money leaves later, when somebody treats the pad as the complete record of a morning, and the pad is the one document from that morning you already know was written under pressure by a busy person.

Run it like a drill

An outage process fits on one page. Numbered outage tickets, so gaps can announce themselves. A hard rule on minimum fields: no truck leaves without customer, product, and job written down, tare be damned. One named owner for reconstruction, whose job is to reconcile the pad against a second source before anything gets keyed. And a marker on every outage ticket in the system, so next quarter you can pull the batch and ask the only question that matters: did we bill everything that left?

You can't schedule the next outage. What gets decided before it arrives is whether it costs an hour of throughput or a morning of revenue.

If you'd rather test the theory than take the argument on faith, send 10-20 sample tickets to hi@termco.ai, including any handwritten ones from your last outage if you kept them, and we'll send back what we find within 48 hours.