Substantive Procedures: Analytics vs. Tests of Details

A believable revenue total still needs evidence. Analytics and tests of details can address the same assertion in different ways; their usefulness depends on the question being tested.

An original hypothetical case

Assume a car wash offers one basic service during a month: 1,000 completed washes at a net price of IDR 20,000 each. Assume no discounts, refunds, add-ons or other revenue streams, with taxes outside the model. The ledger records IDR 22 million. Every input is fictional, not a verified business fact.

The model expectation is 1,000 × IDR 20,000 = IDR 20 million. Recorded revenue exceeds that expectation by IDR 2 million. That is an unexplained model difference—not proof that revenue is overstated or that fraud occurred.

The analytical route

Start with revenue accuracy. In real work, the auditor would evaluate the service-count data, price, period and simplifying assumptions before treating this expectation as useful evidence. A count copied from the same unreliable system as revenue may add little independent challenge.

The question is whether the expectation can identify a potentially material error with enough precision. The acceptable difference is determined in context; this case does not prescribe a universal percentage or rupiah threshold. Differences requiring investigation need supporting evidence, not just a plausible explanation.

The detail route

Suppose management says the difference comes from additional services excluded from the simple model. Select relevant transactions and inspect evidence such as receipts, completed-job records, agreed prices, posting dates and adjustments. In this fictional case, that would test whether those services occurred and whether their recorded values explain the difference.

If the explanation is supported, revise the expectation or conclusion as appropriate. If it is unsupported, investigate further. Do not automatically book IDR 2 million as an adjustment simply because the first model was too simple.

Direction matters

Vouching recorded receipts to completed-job evidence can address occurrence. Tracing completed jobs into the records can address completeness. Both may be useful, but they answer different questions. The main comparison here is how to challenge the accuracy of a revenue amount.

Two limits worth keeping visible

Targeting unusual transactions can test an explanation, but those selections do not automatically represent every transaction. A small total difference can also hide offsetting errors. A matching total is therefore not a blanket conclusion that the underlying activity is correct.

Choose evidence for the assertion

Substantive procedures can involve analytics, details or a combination. Applicable audit requirements and assessed risks constrain that choice. For example, the retrieved ISA-conforming ASA 330 guidance requires details when a significant-risk response consists only of substantive procedures. This is a technical learning comparison, not an assertion that Australian standards govern every Indonesian engagement.

Practice

Add discounts or a second service to the fictional case. Explain which input now needs evidence, how you would rebuild the expectation and which detail test would challenge the proposed explanation. State what your work still would not prove.

Further reading: AUASB ASA 330 and ASA 520. Proposed 2026 IAASB revisions are not treated here as effective requirements.

Beyond the numbers. Inside the business.