The number you pay and bill on is made up
Ask most field operations how they track labor hours and the honest answer is: the tech writes them down. At the end of a long day — or worse, the end of the week — a technician reconstructs from memory how long each job took, rounds to the nearest convenient half-hour, and the office pays payroll and bills customers off that reconstruction. It feels like a record. It's a guess.
And the guess leans one direction. A tech filling in Friday's timesheet from memory doesn't shave time off; they round up, smooth over the gaps, and pad the jobs they're not sure about. You overpay on labor you can't verify and you under-bill the jobs where the real time got lost — because the same fog that inflates payroll hides the jobs that actually ran long. The timesheet isn't dishonest on purpose. It's just memory, and memory is a terrible stopwatch.
The fix isn't a stricter timesheet or a nag to "write it down sooner." It's to stop asking anyone to write it down at all — and capture labor time as an automatic byproduct of work the tech is already doing.
Why handwritten hours fail
When you trace timesheet problems back to causes, the same few show up:
- Reconstruction from memory. Hours logged hours or days after the fact are estimates dressed as records. The longer the gap, the worse the guess.
- Drive time vanishes. Techs log wrench time and forget the windshield time between jobs, so the real cost to deliver — which includes the drive — never makes it onto the books.
- Rounding compounds. Round every job to the nearest half-hour across a week and a crew and the slop adds up to real money, every direction at once.
- No tie to the job. Hours land in a weekly total, not against the specific job, so you can never answer the question that matters: did this job take the time we billed for it?
Notice that none of these are solved by a better paper form. They're solved by removing the human-memory step entirely.
Capture the clock from the job, not the tech
A technician already moves every job through a sequence of states as they work it: scheduled, then en route, then on site, then complete. Each of those transitions is a timestamp the tech is creating anyway — and the span between them is the labor record, captured the moment it happens instead of reconstructed Friday afternoon.
In Hosting Field, the 7-state job FSM stamps those status transitions automatically: when a tech goes en route, when they arrive on site, when they mark the job complete, each is recorded with its timestamp, no separate clock to punch. The on-site minutes and the en-route-to-complete span fall out of the status flow the tech runs anyway. That's the same machinery that powers the per-job trip cost rollup — labor computed from the tech's hourly rate times the recorded minutes, plus fuel and miles — so the hours aren't just tracked, they're already priced. The tech never fills in a timesheet; the timesheet writes itself from the work.
One honest number feeds three systems
The reason this matters beyond payroll is that labor time is the input to three different things that usually run on three different guesses:
- Payroll pays the tech for time actually worked, including the drive, without anyone reconstructing it. The hours you pay are the hours the system recorded.
- Job costing finally sees the true labor cost of each job, because the minutes are attached to the job, not smeared across a week. That's the number that tells you which job types actually make money.
- Invoicing bills from the same recorded time, so the labor line on the invoice matches the labor you paid for — and the job-to-invoice loop closes on real data instead of a Friday estimate.
When all three run on one captured number instead of three separate guesses, the discrepancies that quietly cost you — paid-but-not-billed time, billed-but-disputed time — stop appearing, because there's nothing to reconcile.
Don't weaponize it
There's a trap here identical to the one with GPS tracking: the moment techs believe the time capture exists to catch them slacking, the whole thing turns adversarial. They'll game the status flow — marking "on site" from the driveway, sitting in "complete" — and the data you were trying to clean up gets dirtier than the timesheet it replaced.
Frame it honestly and narrowly. Automatic time capture exists so techs don't have to do timesheets — which is something they actually like, because nobody enjoys reconstructing their week from memory on a Friday. It exists to pay them correctly for the drive time that paper forms swallow, and to cost jobs honestly. Tie it to those wins, never to gotcha discipline, and the same capture that breeds resentment as surveillance becomes the thing that gets people paid right and gets the paperwork off their plate.
What to measure
- Billable hours as a share of paid hours — your technician utilization. The gap between what you pay and what you bill is drive time, idle time, and rework; you can't manage it until you can see it on real numbers.
- Labor estimate-to-actual variance — how close your quoted labor lands to the captured time. Big gaps mean your estimating is running on the old guesses.
- Time-entry lag — how long after the work the hours get recorded. With automatic capture it's zero; if you're still seeing a lag, someone's still reconstructing.
Labor is the biggest cost in field service and, on a handwritten timesheet, the least trustworthy number you have. Capture it off the work the tech already does, attach it to the job, and let payroll, costing, and invoicing all read from the same honest record. The timesheet lie isn't a discipline problem you solve by trying harder — it's a measurement problem you solve by never asking a human to remember in the first place.