All writing
Audit Methodology12 September 20268 min read

Ledger Scrutiny: The Checks Worth Running, and the Four That Cannot Be Sampled

Scrolling a ledger finds what is large. It does not find the month with no entry, the balance on the wrong side, or the party that exists twice.

In short
  • Ledger errors fall into seven kinds — wrong side, missing entries, wrong period, wrong classification, wrong master, a statutory limit crossed, and a suspicious pattern — and each has a different consequence.
  • Completeness, classification and master-hygiene questions cannot be sampled: there is no representative ledger for whether depreciation was posted or whether a party exists twice.
  • An exception is only evidence once it is routed — to a disallowance, a clause of the tax audit report, a Schedule III disclosure or a management letter.
A desk with papers, spectacles, a calculator and office suppliesPhotograph: Cht Gsml / Unsplash

"Ledger scrutiny" is the part of an audit programme most likely to mean whatever the person holding the file thinks it means. Usually it means scrolling — opening each ledger, looking at the big numbers, and stopping when nothing looks odd.

The trouble with scrolling is that it finds what is large, and almost nothing else. The findings that cost a client money are usually small, or are not a number at all: a balance on the wrong side, a month with no entry, a party that exists twice.

What follows is the same work, written as tests.

Seven things a ledger can be wrong about

Every check worth running belongs to one of these. The value of the grouping is that each has a different consequence, so each has a different owner.

1. The balance is on the wrong side. A debtor with a credit balance is an advance from a customer, and under Schedule III it belongs on the other side of the balance sheet, not netted off. A creditor with a debit balance is the same problem in reverse. A bank ledger that is overdrawn is a borrowing. None of these is an error in the ledger; each is an error in the financial statements if carried through.

2. Something that should exist does not. This is the largest and least examined category, because nothing on the screen draws attention to an entry that was never made. Depreciation not posted. Interest on a term loan accrued for eleven months. Rent income for a property recorded in ten months of the year. The provision for tax, the deferred tax, gratuity, leave encashment, bonus. The closing stock entry. The profit or loss on an asset sold. Amortisation on intangibles. Each of these is found by asking "is it there?", never by looking at what is.

3. The period is wrong. Entries dated outside the financial year. Prepaid insurance or an annual maintenance contract charged in full. March salary and commission that belong to March but were paid — and booked — in April. Year-end foreign exchange restatement.

4. The classification is wrong. Capital expenditure sitting in a revenue head. Personal expenses that should be drawings. A profit and loss item parked in the balance sheet. Ledgers the mapping could not classify at all — which have to be surfaced rather than defaulted, because a silent default is how a misclassification survives into the financial statements.

5. The master is wrong. The same party as two ledgers with slightly different names, so neither balance is complete and the ageing of both is wrong. A vendor with no PAN or GSTIN. Suspense and temporary accounts still carrying a balance at the year end.

6. A statutory line has been crossed. Cash payments above the section 40A(3) limit. Loans, deposits and receipts against sections 269SS, 269T and 269ST. Partner remuneration and interest against the section 40(b) limits, and now section 194T on partner payments. CSR under section 135. Each of these has a clause of the tax audit report waiting for it.

7. The pattern is wrong. Round-figure journals. Manual journals clustered in the last week of the year. A provision created and reversed in the next period. Voucher numbers out of sequence. A party appearing as both debtor and creditor. Duplicate payments to a creditor. These are not errors on their face — they are reasons to look.

Where sampling stops working

Categories 1, 2, 4 and 5 cannot be sampled. There is no representative ledger for "is depreciation posted", and a sample of forty vouchers will not tell you that a party master exists twice. They are population questions, and they have to be asked of the whole population or not at all.

Categories 3, 6 and 7 can be sampled, but the sample has to be chosen by the test rather than at random: every voucher above the cash limit, every entry in the last week of March, every round-figure journal.

The reason ledger scrutiny gets skipped is not that it is hard. It is that done by hand, on a population, it does not fit in the time available.

The programme

A one-page programme with the checks, the consequence each carries, and columns for the exception found and where it went — the disallowance schedule, a clause of the tax audit report, a Schedule III disclosure, or a management letter point.

Download the ledger scrutiny programme (.xlsx) — free, no sign-up.

Where each exception ends up

The thing that turns scrutiny into audit evidence is routing. A finding that does not reach a clause, a disclosure or a letter is a note in a margin.

  • Wrong-side balances → the Schedule III grouping, and the ageing that depends on it.
  • Missing accruals and provisions → the financial statements, and a management letter point about the closing process.
  • Cash limits and 40(b) → the disallowance schedule, and clauses 21 and 31 of the tax audit report. The clause-wise routing is in the tax audit checklist.
  • Creditors outstanding long enough to raise a cessation question → section 41(1), and the ageing in the Schedule III disclosures.
  • Related party ledgers → the note and the register, covered in related party transactions.
  • Pattern findings → the fraud discussion required by the standards, documented whether or not anything came of it.

How Audcrix runs it

Audcrix runs 51 accounting checks and 30 named fraud indicators across the whole population of ledgers and vouchers, once the books are read from Tally, rather than on a sample.

  • Each check states the consequence on the row — the section or the standard it comes from — and opens onto the vouchers behind it, so the figure is never the only thing on screen.
  • The completeness checks are the ones that pay for themselves: depreciation, interest accruals, provisions, closing stock, amortisation, month-by-month rental income, prepaid splits.
  • Ledgers that cannot be classified are reported as unclassified rather than defaulted into a group, because a silent default is the failure mode that survives review.
  • Where evidence for a conclusion is not present — no certificate, no confirmation — the row is flagged as unresolved rather than assumed either way.

None of that replaces the judgement about whether a debtor is recoverable or a dispute is genuine. It replaces the scrolling.

Questions this answers

What is ledger scrutiny in an audit?

Examining the ledgers for errors of side, completeness, period, classification, master hygiene, statutory limits and pattern — as defined tests with a stated consequence, rather than reading through balances.

Which ledger checks cannot be done on a sample?

Completeness, classification, wrong-side balances and master hygiene. There is no representative ledger for whether depreciation was posted, and a sample will not show that the same party exists as two ledgers.

What does a debtor with a credit balance mean?

It is generally an advance from a customer, which under Schedule III belongs on the liabilities side rather than netted against trade receivables.

Which ledger findings feed the tax audit report?

Cash payments above the section 40A(3) limit and loans, deposits and receipts under sections 269SS, 269T and 269ST feed clauses 21 and 31, and partner remuneration and interest under section 40(b) feed the disallowance schedule.

How many checks does Audcrix run on the ledgers?

51 accounting checks and 30 named fraud indicators, run across the whole population of ledgers and vouchers rather than on a sample, each stating the section or standard it comes from and opening onto its vouchers.