The Reconciliations That Actually Catch Things
Why this matters
A reconciliation is the only routine in your books that can prove anything. Every other report is an opinion assembled from whatever somebody typed in; a reconciliation compares your record against a record kept by an outside party who has no interest in flattering you - a bank, a card issuer, a lender, a payroll processor, a tax authority - and forces the two to agree. Errors that survive reconciliation run for a year and then surface at tax time, in a loan application, or in a buyer's due diligence, by which point every downstream number built on them is also wrong.
The trap is not skipping reconciliations. It is doing a dozen of them shallowly and clearing the differences by plugging. A shop that reconciles everything and forces the last few items into a catch-all account has trained itself to trust a number that has never actually been tested.
What a reconciliation proves, and the four words you need
Subledger: the detailed list behind a summary account. Your open-invoice list is the accounts receivable subledger; the single AR figure on the balance sheet is the control account. They should equal each other exactly, always, with no rounding tolerance.
Clearing account: a holding account money passes through, not a place it lives. Undeposited funds and merchant-processor clearing are the two you will meet. A healthy clearing account empties out constantly; a clearing account with a growing balance is a symptom, not a balance.
Plug: an entry made to force a reconciliation to agree without knowing why it disagreed. A plug converts a known error into an unknown one and is the single most damaging habit in small-shop bookkeeping.
The rule, per account, per period, with no size exception: a reconciliation is finished when the unexplained difference is zero. Not small, not immaterial, zero. If a difference survives the close, it does not get plugged - it becomes a written open item with a named owner and a due date, and it is the first thing reviewed next month.
The seven that earn their time
These are the reconciliations where the outside record exists, is authoritative, and catches something the shop cannot catch any other way.
| Reconciliation | Outside record | What only this catches | Cadence |
|---|---|---|---|
| Operating bank account | Bank statement | Unrecorded fees, duplicate or missing deposits, checks that never cleared, unauthorized withdrawals | Monthly, plus a weekly balance glance |
| Every credit card and line of credit | Issuer statement | Personal spend on a business card, subscriptions nobody remembers approving, interest and fees never booked | Monthly, each card separately |
| Merchant clearing and undeposited funds | Processor settlement report | Gross-versus-net recording errors, a batch recorded twice, refunds and chargebacks never entered | Monthly, and any item aged past 5 business days gets chased individually |
| AR subledger to the AR control account | Your own open-invoice list | Payments applied to the wrong customer, credit memos issued but never posted, invoices deleted rather than voided | Monthly |
| Payroll liabilities | Payroll provider reports and tax notices | Withholdings collected but not remitted, an employee paid outside the system, an accrual that never reverses | Each payroll cycle, reconciled monthly |
| Sales tax payable | The returns you actually filed | Tax collected at one rate and remitted at another, exempt sales taxed, a jurisdiction filed late | Every filing period |
| Loan and finance balances | Lender statement | The whole payment expensed instead of split between interest and principal | Monthly, or quarterly on a fixed-term note |
The loan row is the one owners skip and the one that most reliably distorts profit. Recording an entire loan payment as an expense understates profit by the principal portion every single month, and on an amortizing note the principal share grows over the term, so the distortion silently increases year over year.
What deliberately comes off the list, and why
This section is the point of the card. Time spent on the wrong reconciliation is time not spent on the seven above.
Full physical inventory of parts, monthly. For most service shops, truck stock and shelf stock churn too fast and cost too little per line for a monthly wall-to-wall count to pay. Cycle count instead: pick the ten highest-value or highest-theft-risk items and count those every month, count everything once a year. The exception that flips this: if parts are a large share of your cost of goods sold and you are seeing unexplained shrink, a full count becomes the diagnostic, not the routine.
Accounts payable subledger, monthly, when you pay on receipt. If the shop pays bills as they arrive and carries essentially no vendor balances, the AP subledger reconciliation confirms what you already know. Reconcile AP monthly the moment you start carrying vendor terms, and reconcile the two or three largest supplier accounts against their statements regardless, because supplier statement errors are common and they are the vendors who can hurt you.
Equity accounts, monthly. Owner contributions, draws and retained earnings do not need a monthly outside comparison because there is no outside record. Review equity once a year with your accountant, where it belongs.
Every expense account against receipts, monthly. This is not a reconciliation at all, there is no counterparty record, and it eats hours. Sample instead: pull five to eight transactions a month across different accounts and confirm each has a real supporting document. That tests whether the documentation habit is working, which is the actual thing you want to know.
Job costs against the P&L, line by line. This matters enormously and it is not a reconciliation. It is job costing review, it uses different evidence, and treating it as a tie-out means you will do it as a clerical exercise instead of a management one.
Worked example: a clean bank rec hiding a real error
A shop closes a month. The operating account reconciles to zero difference on the first pass, which the owner takes as a sign the month is clean. The merchant clearing account is the one the bookkeeper flags.
The clearing account holds 11 open items. Nine of them are under 3 days old, which is ordinary settlement float and needs no work. Two are aged past 40 days. Under the stated rule - any clearing item older than 5 business days is chased individually, per item, regardless of size - both get pulled.
Item one is a card batch that appears twice: once when the office entered the day's payments by hand, and again when the bank feed imported the settlement deposit. Revenue for that day was recorded twice and the second copy never cleared because it never existed as a second deposit.
Item two is a customer refund the office issued through the processor and never recorded in the books at all. The money left the merchant account; the ledger still shows the original sale in full.
Both errors push revenue the same direction, up. Sized against that month's card volume, the duplicate batch is about 2% of card revenue and the unrecorded refund is under 1%, so combined they overstate card revenue by roughly 3% for the month, on a base of that month's card volume only.
Three percent sounds survivable, and if it were a one-month rounding issue it would be. It is not. The duplicate-entry mechanism is procedural: the office enters payments by hand and the feed also imports them, so the same collision happens every month somebody keys a batch before the feed lands. Left alone for a year it inflates the revenue line, inflates the sales tax the shop believes it owes on card sales, and produces a receivables balance that will never collect because the second copy of the batch sits against invoices that were already paid.
What the shop changed. One person owns card-payment entry, and that person enters payments only from the processor's settlement report, never from the day's paper. The bank feed is set to match, not to create. That removes the mechanism rather than correcting one month.
What would flip the recommendation. If the shop had a single low-volume card terminal and no bank feed, hand entry from the day's receipts would be the right method and the control would be a weekly tie of terminal totals to deposits instead. The rule is not "never enter by hand"; it is "one source of truth per transaction type."
The failure mode if you get this wrong. The tempting fix is a journal entry that zeroes the clearing account at month end. It reconciles, the report looks finished, and the two real errors are now permanently unfindable because the evidence trail was closed. That is a plug, and it is how a shop ends up unable to explain its own revenue two years later.
How to verify you got this right
- The zero test. Open last month's reconciliation reports. Every one should show an unexplained difference of exactly zero, and any account that does not should have a written open item with a name and a date next to it. A difference described as "small" is an unfinished reconciliation.
- The plug test. Search the period for journal entries hitting a clearing, suspense or miscellaneous account with no supporting document. Any hit is a plug and needs to be re-opened, not re-posted.
- The aging test. Print the open items in undeposited funds and merchant clearing. Every item should be younger than the settlement window for your processor, typically a few business days. An item aged past 30 days is almost never float.
- The subledger test. Pull the AR subledger total and the AR figure on the balance sheet on the same date. If they differ by any amount, the difference is the finding, and the usual causes are a deleted invoice or a payment applied to the wrong customer.
- The independence test. The person who receives money should not be the only person who reconciles the account that receives it. In a shop too small to split those duties, the owner reviews and initials the completed reconciliation, which is a real control and takes minutes.
References
- Generally Accepted Accounting Principles (GAAP), control account and subsidiary ledger agreement
- See related: When the Bank Balance Lies: Reconcile Discipline, Chart of Accounts Design for a Service Business, Month-End Close Checklist for a Service Business, Reading Your Balance Sheet Basics