What Is Account Reconciliation For if the Books Balance?

What is account reconciliation? It compares a balance in the books with an independent record of the same account, usually a bank or processor statement, and explains every difference. The agreement proves little. Most differences are timing, and each payment rail sets how long timing can last; a difference older than that is an error or a loss.

In the Association of Certified Fraud Examiners’ 2026 study of 2,402 cases, account reconciliation was how 5% of frauds were first detected. Those frauds ran a median 8 months before discovery, against 12 for frauds exposed by a tip. The same study found that 7% of fraudsters had “forced or altered account reconciliations” to cover their tracks (ACFE, 2026). Each case was the largest a responding examiner had investigated, so the figures describe frauds that were caught and say nothing about how common fraud is.

What each common reconciliation checks, and what a stale difference usually means

Account in the books	Compared with	Normal timing differences	A difference that outlives its timing usually means
Operating bank account	The bank statement	Deposits after the cut-off, uncashed checks, payments still settling	A deposit never made, a payment drawn twice, an unrecorded fee
Payment processor clearing account	The processor’s payout and balance reports	Payouts in transit	Refunds, disputes or fees netted from payouts and never booked
Payroll liabilities	The payroll provider’s reports and tax filings	Taxes withheld but not yet remitted	Payroll tax unpaid or posted to the wrong account
Accounts receivable	The customer ledger and customer statements	Payments received but not yet applied	Cash applied to the wrong invoice, or invoices nobody will pay
Prepaids, accruals and capitalised costs	Internal schedules and contracts; no outside record exists	Amortisation not yet posted	Expenses parked on the balance sheet
Suspense and clearing accounts	Nothing: the balance should return to zero	Items awaiting classification for a day or two	Errors or losses nobody has claimed

Source: CX Cash synthesis; the ranking of outside records follows PCAOB AS 1105.

What is account reconciliation, and why is a balanced ledger not enough?

Any answer to what is account reconciliation starts with a check the books already run on themselves, and that check is weak. “A trial balance only checks the sum of debits against the sum of credits”, so it catches any error that leaves one side short. It passes every error that keeps both sides equal: a transaction left out entirely, a wrong amount entered on both sides, a cost posted to the wrong kind of account (Trial balance, Wikipedia).

Reconciliation adds information the ledger cannot produce about itself. Auditing standards rank evidence by where it comes from: “Evidence obtained from a knowledgeable source that is independent of the company is more reliable than evidence obtained only from internal company sources” (PCAOB AS 1105). For cash, the bank statement is that source. Prepaid expenses and capitalised costs have no bank, so they can be reconciled only to schedules the company keeps itself, and that is where a cost posted to an asset account can sit unnoticed.

Commercial Vehicle Group, a publicly traded manufacturer, found one in 2020. While preparing its 2019 annual report it noticed a potential overstatement of a prepaid production tooling account. An investigation traced it to “a former employee preparing manual journal entries to understate cost of revenues by improperly capitalizing certain manufacturing expenses”. The company told investors to stop relying on its 2018 accounts and six quarterly reports. It named two material weaknesses: controls over “balance sheet account reconciliations and over review of manual journal entries” (Commercial Vehicle Group, Form 8-K, 2020). Every one of those journal entries balanced.

French bookkeepers have a word for the matching step: lettrage, from the practice of marking entries that settle one another with a shared code, often a letter. Its stated purpose is to pair the entries and to establish which ones remain unpaired (our translation; Lettrage comptable, Wikipédia). The unpaired entries are the output, so the practical answer to what is account reconciliation is a dated list of what did not match. A reconciliation that ends with a total that agrees and no such list has done the bookkeeping and skipped the finding.

Why do the bank balance and the book balance differ?

Mostly because money takes time to move, and each payment rail has written rules for how long it can sit between the payer’s books and the payee’s bank.

A deposit made after the bank’s cut-off, which Regulation CC lets a bank set at 2:00 p.m. or later, may be treated as made on the next banking day (12 CFR 229.19). Nacha, which writes the ACH rules, says credits settle the same day, the next banking day or in two banking days. It adds that “ACH debits cannot have a settlement date that is more than one banking day into the future”, and estimates that 80% of ACH payments settle in one banking day or less (Nacha). Stripe pays out card receipts on a schedule set by its settlement timing, standard T+2 in the United States (Stripe). The Federal Reserve’s Fedwire service handles same-day transfers with final settlement (Federal Reserve Financial Services).

Checks a company writes run on the longest clock. Under the Uniform Commercial Code, a bank “is under no obligation to a customer having a checking account to pay a check, other than a certified check, which is presented more than six months after its date” (UCC § 4-404).

Those rules give every reconciling item a time limit: the longest a genuine timing difference can last on the rail that produced it. Inside the limit, a difference is timing and needs only a date. Past it, the difference has some other cause, and the reconciliation should name it.

How long each common reconciling item can legitimately stay open

Reconciling item	Why the bank does not show it yet	Rule or published schedule	Time limit	Past the limit, look for
Deposit in transit	Made after the cut-off or on a non-banking day	Regulation CC, 12 CFR 229.19	Next banking day	A deposit never made, or made to another account
Incoming ACH credit	Settlement date chosen by the sender	Nacha: same day, next day or 2 banking days	2 banking days	A payment the customer never sent
Outgoing ACH debit	Settlement date chosen by the originator	Nacha: at most 1 banking day ahead	1 banking day	A debit drawn twice, or never authorised
Card-processor payout	The processor’s settlement period	Stripe: standard T+2 in the US	2 business days, plus the bank’s posting	Refunds, disputes or fees netted from the payout
Wire	Same-day settlement	Fedwire	Same day	A wire recorded but never sent, or sent twice
Check written, not yet cashed	The payee has not deposited it	UCC § 4-404	6 months	A void check, or a payee still owed the money

Sources: as linked above. The time limits are CX Cash’s reading of each rule.

How do you tell a timing difference from an error?

Date it and compare its age with its time limit. The example company is hypothetical, a seed-stage software business closing its books on Wednesday 30 September 2026. Its ledger shows $428,715 in the operating account and the bank statement shows $412,380. Eight items explain the $16,335 between them, and with all eight listed the reconciliation agrees to the cent.

A reconciliation that agrees to the cent, aged against each item’s time limit

Item	Books minus bank	Age on 30 Sep	Time limit	Verdict
Check deposited at 4:10 p.m. on 30 Sep, after a 2:00 p.m. cut-off	+$6,000	Same day	Next banking day	Timing: due Thursday 1 Oct
Processor payout for card charges captured 29 Sep	+$8,400	1 business day	2 business days	Timing: due Thursday 1 Oct
Customer check recorded as deposited on 18 Sep	+$4,500	8 banking days	Next banking day	Past the limit: find the check
Check #1042 to a contractor, written 15 Sep	−$3,200	15 days	6 months	Timing
Check #1007, written 12 Feb	−$1,850	230 days	6 months	Past the limit: void it and settle with the payee
Interest credited by the bank	−$310	Same day	Record now	Book it
Monthly account fee	+$45	Same day	Record now	Book it
Vendor ACH debit on 9 Sep, a second draw on one invoice	+$2,750	15 banking days	1 banking day	Past the limit: reclaim it from the vendor
Total	+$16,335

Source: hypothetical company, with time limits from the table above; CX Cash calculation.

Labelled “timing”, all eight items close the reconciliation; aged, three of them fail. Of the $16,335 gap, $11,200 is timing and $265 is bank interest, net of a fee, waiting to be booked. $7,250 of it, 44% of the gap, is cash the books record and the bank does not hold (our arithmetic). The $1,850 check is the opposite case: money still in the bank that the company owes to someone who has not cashed it, and that the bank may now refuse to pay.

The duplicate debit shows why the age matters more than the total. Aged, it is 15 banking days past a limit of one, and the vendor has had $2,750 of the company’s cash for three weeks. A reconciliation that dates its leftovers turns that line into a phone call in the first week.

What happens to a difference nobody explains?

It can grow, as Barings showed at the extreme. In 1992 Barings, then the United Kingdom’s oldest merchant bank, sent Nick Leeson to Singapore to run both the trading desk and the back office that settled its trades. He hid losses in an error account numbered 88888, and by his own later account he crossed into crime when he forgot to reconcile a discrepancy of 500 contracts that cost the bank $1.7 million (Nick Leeson, Wikipedia). The account’s losses exceeded £2 million at the end of 1992, then reached £23 million in late 1993 and £208 million by the end of 1994. They ended at £827 million, twice the bank’s available trading capital, and Barings was declared insolvent on 26 February 1995 (Barings Bank, Wikipedia).

A startup’s bank account is not a derivatives book, but the structural fault travels: one person kept both records. Barings and the ACFE’s figure of 7% suggest that a reconciliation catches losses sooner when someone other than the person who made the entries prepares or reviews it, and becomes one more place to hide when the same person does both.

Can the record you reconcile against be trusted?

Usually, but independence can fail at the source. Satyam Computer Services, an Indian IT company listed in New York, overstated its revenue, income and cash by more than $1 billion over five years. According to the SEC, employees produced bogus bank statements, and the result was “more than $1 billion in fictitious cash and cash-related balances, representing half the company’s total assets” (SEC, Litigation Release 21915, 2011). PwC’s Indian affiliates “routinely relinquished control of the delivery and receipt of cash confirmations entirely to their audit clients” (SEC, 2011). Satyam’s auditors let the company that had invented the cash deliver the letters confirming it.

Few startups will meet a forged statement; the everyday version ships with accounting software. Bank feeds import each transaction from the bank and bank rules categorise it. In QuickBooks Online a rule can also post it straight to the ledger, and Intuit’s own help page warns that “if you leave it on you won’t get a chance to review transactions before QuickBooks adds them to your books” (Intuit). An entry created from a bank line always matches that bank line. Had the vendor’s second draw in the worked reconciliation met such a rule, it would have been booked as software spending and the reconciliation would have shown nothing. AS 1105 has a sentence for this: “Obtaining more of the same type of audit evidence, however, cannot compensate for the poor quality of that evidence.”

For entries created from the feed, the independent check moves to the vendor’s invoice, the contract and the agreed price. The bank reconciliation still proves that every bank line reached the ledger; it no longer proves that each payment should have happened.

Does a startup need formal reconciliation controls?

It needs the reconciliation more than the audit of it. Among listed companies, the ones with weak controls share traits with startups. Jeffrey Doyle, Weili Ge and Sarah McVay studied 779 firms that disclosed material weaknesses from August 2002 to August 2005; weaknesses were more likely in firms that are “smaller, younger, financially weaker, more complex, growing rapidly, or undergoing restructuring” (Doyle, Ge & McVay, 2007). Ge and McVay’s earlier sample of 261 companies listed inappropriate account reconciliation among the common causes (Ge & McVay, 2005). Private startups disclose nothing comparable, so this evidence about them is indirect.

A material weakness, under PCAOB AS 2201, means “a reasonable possibility that a material misstatement of the company’s annual or interim financial statements will not be prevented or detected on a timely basis” (PCAOB AS 2201). For listed firms it has a price: Hollis Ashbaugh-Skaife and three co-authors found that when auditors confirmed a change in control quality, the cost of equity moved by amounts that “range from 50 to 150 basis points” (Ashbaugh-Skaife et al., 2009).

The SEC has argued the other side for small companies. In 2020 it exempted smaller reporting companies with annual revenues under $100 million from the auditor’s attestation on internal control. Issuers below that line had restatement rates three to nine percentage points lower, and their financial statements “may be less critical to assessing their valuation given, for example, the relative importance of their future prospects”. The Commission put the average saving at about $210,000 a year, roughly $110,000 of it in audit fees (SEC, Release 34-88365, 2020). The SEC is right about the attestation: a company valued on its prospects gains little from a six-figure audit of its controls.

The same release left those companies obliged to “establish and maintain ICFR and have management assess the effectiveness of ICFR”, so it removed the outside check and kept the controls. A monthly bank reconciliation at a seed-stage company probably takes a few hours, and the balance it tests, cash, is the one a young company’s prospects depend on.

What is account reconciliation, in the end? A total that agrees can be produced by mislabelling; what the reconciliation leaves over, dated and owned, is the finding. In the worked example, three of the eight items had outlived their time limits, and two of them were $7,250 of cash the books recorded and the bank did not hold.

References

• Association of Certified Fraud Examiners (2026). Occupational Fraud 2026: A Report to the Nations.
• Ashbaugh-Skaife, H., Collins, D. W., Kinney, W. R. and LaFond, R. (2009). The effect of SOX internal control deficiencies on firm risk and cost of equity. Journal of Accounting Research 47(1), 1–43.
• Commercial Vehicle Group, Inc. (2020). Form 8-K, Item 4.02, filed 16 March 2020.
• Doyle, J. T., Ge, W. and McVay, S. E. (2007). Determinants of weaknesses in internal control over financial reporting. Journal of Accounting and Economics 44, 193–223.
• Federal Reserve Financial Services. Fedwire Funds Service.
• Ge, W. and McVay, S. E. (2005). The disclosure of material weaknesses in internal control after the Sarbanes-Oxley Act. Accounting Horizons 19(3), 137–158.
• Intuit. Set up bank rules to categorize online banking transactions in QuickBooks Online (help article).
• Nacha. How ACH payments work.
• PCAOB. AS 1105: Audit Evidence; AS 2201: An Audit of Internal Control Over Financial Reporting That Is Integrated with An Audit of Financial Statements.
• Regulation CC, 12 CFR 229.19.
• SEC (2011). Litigation Release No. 21915, SEC v. Satyam Computer Services Limited; press release 2011-82, SEC Charges India-Based Affiliates of PwC for Role in Satyam Accounting Fraud.
• SEC (2020). Accelerated Filer and Large Accelerated Filer Definitions, Release No. 34-88365.
• Stripe. Receive payouts (documentation).
• Uniform Commercial Code, § 4-404 (Legal Information Institute, Cornell Law School).
• Wikipedia. Barings Bank; Nick Leeson; Trial balance. Wikipédia. Lettrage comptable.

The Barings figures come from Wikipedia’s summary of contemporary reporting and of Leeson’s own account. The ACFE’s figures describe frauds that were caught, and the control-weakness studies describe listed companies, so both are indirect evidence about startups. Figures without a citation are worked out from the cited statistics and rules, and the seed-stage company in the worked reconciliation is hypothetical. Translations from French are ours.