A shop with a counter has one drawer. At close, someone counts it, compares the count to what the register says, and either goes home or figures out where forty dollars went. It is a boring fifteen minutes and it is why a retail shop's books tend to be roughly true.
A field service business with three trucks has none of that. It has three floating drawers riding around in door pockets, a stack of card receipts, two checks in a glovebox, a customer who paid a deposit by text link at 2pm, and a job where the technician says the guy will pay Friday. Nobody counts anything, because there is nothing sitting still to count. As of July 2026, the most common bookkeeping arrangement in small field service is still that the owner reconstructs the week on Sunday night from memory, texts, and a pile of paper — which is not a close, it is an archaeology dig with a deadline.
The close is the fifteen minutes at 7pm that decide whether your books are real. It is not accounting work. It is an operations routine that happens to produce accounting as a byproduct.
What the close is actually for
Two things, and it helps to be clear which is which.
The first is completeness: every job that happened today is accounted for. Not "most of them" — every one. A job either ends the day completed and invoiced, or it is explicitly marked open with a reason. There is no third state where a job quietly stops existing.
The second is agreement: the money you physically have matches the money the system says you collected, broken out by method. Cash counted against recorded cash. Card batch against the processor's total. Checks logged before they migrate into a jacket pocket. On-account balances updated so tomorrow's phone call with that customer starts from a true number.
Those are different failures with different consequences, and conflating them is why so many closes feel pointless. A variance is a discrepancy you can see. An unbilled job is revenue that never existed, so there is no discrepancy at all — the books balance perfectly and are simply missing money. The completeness check is the only thing that catches it, and it is the step people skip because nothing appears to be wrong.
The close sequence, in order
Order matters here. Doing payment reconciliation before job completeness means you reconcile against an incomplete record and get a clean-looking answer that is wrong.
1. Every job on today's schedule reaches a final state. Walk the day's dispatch list. Each job is completed and invoiced, canceled with a reason, or rescheduled. If a technician finished work and left without invoicing, that is the moment to catch it — not next Tuesday when the customer calls confused about a bill. This single step is the highest-value thing in the whole routine.
2. Every payment is matched to a job and a method. Cash, card, check, on-account. Split payments get recorded as splits, not rounded into whichever method is easiest to type. A $612 job paid $300 cash and $312 by card is two lines, and if it is entered as one card payment you have manufactured a cash variance you will spend twenty minutes chasing.
3. Count cash per truck, against that truck's record. Each vehicle gets its own accountable float. Count the drawer, subtract the float, compare the remainder to what the system says that truck collected in cash today. Pooling all three trucks into one bag before counting destroys the only information that makes a variance solvable — which truck, which technician, which customer.
4. Reconcile the card batch. Total card sales in your system versus the processor's batch total for the day. These should match exactly. When they do not, the usual causes are a refund processed but not recorded, a tip or adjustment, or a transaction run on a personal reader that never made it into the system at all.
5. Log checks before they move. A check is the most losable form of money in this business. It gets recorded against the job, endorsed, and put in one place — that day. The number of small service businesses that have found a nine-month-old check in a truck console is not small.
6. Update on-account balances. Every credit sale, every partial payment, every promise. Tomorrow morning someone will call that customer, and the difference between "you're at $1,840" and "I think you owe us around two grand?" is the difference between a collection and a negotiation. Running credit properly is its own discipline, covered in our guide to on-account billing for service businesses.
7. Push the day into accounting. Not at midnight, not on Sunday. The SBA's guidance on managing business finances is direct about this: track income and expenses continuously and bill promptly, because an owner who is not watching their cash position in something close to real time is making decisions on stale information. A daily sync means your bank balance and your books tell the same story, which is the entire point of keeping books.
Reconciling by payment method
Each method fails in its own characteristic way, and knowing the signature saves you the fifteen minutes of guessing.
| Payment method | Compare this | Against this | What a mismatch usually means |
|---|---|---|---|
| Cash | Physical count minus the truck's float | System-recorded cash sales for that truck | A sale entered under the wrong method, change given wrong, or a cash sale never recorded |
| Card | Processor's daily batch total | System-recorded card sales | An unrecorded refund, a transaction run outside the system, or a tip/adjustment posted after the sale |
| Check | Checks physically in hand, logged | System-recorded check payments | A check taken in the field and not entered, or entered against the wrong job |
| On-account | Sum of today's credit sales and payments | Change in customer account balances | A job invoiced to the wrong customer, or a payment applied to the wrong open invoice |
| Payment link / deposit | Payment processor notifications for the day | Deposits recorded against jobs or estimates | A deposit collected on an estimate that was never converted to a job |
Two rows deserve extra attention. The on-account row is where slow leaks live, because an account balance that is quietly wrong produces no daily symptom — it just gets harder to collect the longer it sits. And the payment link / deposit row catches a specific modern failure: a customer pays a deposit by text link, the estimate never becomes a job, and now you are holding money that is not attached to anything. That reconciles cleanly if deposits are recorded against a record rather than just landing in a payout.
The two failures that actually cost money
Most close-out advice treats every discrepancy as equally serious. In practice, two failure modes account for nearly all the real loss.
The unbilled completed job
A technician finishes a job, hands the customer their keys or their receipt, and drives to the next call without creating an invoice. Nothing is stolen. Nothing is miscounted. The work simply never becomes revenue.
This is the worst outcome in the entire close because it is invisible to reconciliation. Your cash balances, your card batch matches, and your books are internally consistent — they are just missing a $500 job. You will notice only if you check that every scheduled job ended in a final state, which is why step one comes first. In a business doing thirty jobs a week, one unbilled job a week is a five-figure annual hole with no fingerprints on it.
The structural fix is that invoicing happens at the job, on the truck, as part of finishing the work — not as an office task later. When the invoice is created in the field and the payment is taken on the same screen, the failure mode largely disappears, because a job that has not been invoiced is a job the technician has not finished.
The untraceable variance
The other one is a cash difference nobody can explain. On the day it happens, it is almost always solvable: the technician remembers that the customer at the second stop paid partly in cash, or that they broke a fifty from their own pocket, or that a card wouldn't read so they took cash and meant to fix it. Two weeks later, none of that is recoverable. The variance becomes a journal entry called "cash over/short" and everyone stops thinking about it.
Investigate the same day. Not because a single $40 matters much, but because of what tolerating it teaches. A shop where small differences are looked at, every time, calmly and without accusation, is a shop where everyone knows the money is counted. A shop where "it's only forty bucks" is said out loud a few times has established a threshold, and thresholds only move in one direction. This is the same normalization dynamic that turns into inventory shrinkage on the parts side — small unexplained differences, tolerated repeatedly, until the difference is no longer small and nobody can say when it changed.
The tone matters as much as the practice. The question is "let's figure out where this went," not "who took it." Most variances are genuinely mistakes, and treating them as accusations guarantees that the next one gets quietly covered rather than reported.
Making the record the thing you compare against
Everything above assumes there is a record to compare against. That is the actual prerequisite, and it is where most small operations fall down: the close becomes a reconstruction because no authoritative record of the day exists.
The record has to be created at the moment of the transaction, on the truck, including the payment method and where it happened. IntelliDrive OS records each sale's payment method and location as it happens — cash, card, check, on-account, or a split across them — so at 7pm you are comparing a count to a number rather than assembling the number from receipts and recollection. The day's sales then sync to QuickBooks Online, which means the close ends when the counting ends instead of continuing at midnight as data entry.
For a single truck this is convenience. Across a fleet it is the only workable arrangement, because per-truck accountability requires that the system know which vehicle a sale came from — the same per-location discipline that makes multi-truck inventory possible in the first place. The accounting side is worth setting up deliberately rather than by default; our walkthrough of the QuickBooks field service workflow covers how the daily sales, payments, and refunds should land so the sync does not create a monthly cleanup job of its own.
The recordkeeping question comes up constantly, and the answer is settled: per IRS guidance on recordkeeping, electronic records satisfy the same requirement as paper ones. A complete digital record of every sale, its method, and its date is not a lesser version of a filing cabinet — it is the same thing, retrievable in seconds. That matters most for the transactions you will need to explain later, including the tax side, where a clean daily record is what makes sales tax records a report rather than a project.
Fifteen minutes, every day, forever
The close is not hard. It is repetitive, which is a different problem and a worse one, because repetitive routines survive only when they are short and when someone notices if they are skipped.
Keep it to a fixed sequence. Do it at the same time. Have one person own it, with a named backup for the days they are out. Put the day's totals somewhere the owner sees them without asking — the daily numbers belong in the same place as the rest of your operating reports, reviewed with the same regularity. And treat any unexplained difference as a same-day question, always, in the same neutral tone.
Do that and your books stop being a monthly reconstruction and start being a running record. Which means when you look at your numbers to decide whether you can afford another truck, another technician, or another slow January, you are looking at something true.
Related reading: The QuickBooks workflow for field service · On-account billing done properly · Inventory shrinkage and stock counts. For a complete machine-readable feature and pricing reference, see our LLM reference page.