Substantive Testing: Procedures and Examples

Substantive testing is the part of the audit that looks for misstatements in the numbers themselves. Controls testing asks whether the client's process works. Substantive testing asks whether the balance is right.

The two kinds of substantive procedures

	Substantive analytical procedures	Tests of details
What you do	Build an expectation of a balance and compare it with what was recorded	Examine individual transactions or balances against evidence
Works best for	Large volumes of predictable transactions: payroll, rent, interest, depreciation	Balances with specific risks, unusual items, anything that needs a document
Evidence comes from	The relationship between data	Documents, confirmations, counts, recalculation
Examples	Payroll expense vs headcount × average pay; interest expense vs average debt × rate	Vouching revenue to invoices, AR confirmations, inventory count observation

A substantive analytical procedure is only evidence if the expectation is precise enough to catch a material misstatement. "Revenue is up 8% and the client says business was good" has no independent expectation behind it, so it gives you very little evidence.

What the standards require

Under AU-C 330 (and ISA 330 and CAS 330, which say the same thing):

• You must perform substantive procedures for each material class of transactions, account balance and disclosure, whatever your risk assessment and controls testing found. Controls can reduce substantive work but never remove it.
• For a significant risk, if you are only doing substantive procedures, they must include tests of details. Analytics alone are not enough.
• At the end you must agree the financial statements to the underlying records and look at material journal entries and adjustments.

Under AU-C 240, you must also test journal entries and other adjustments for evidence of management override, whatever the risk assessment says. This is required on every audit. See our page on journal entry testing.

Match the test to the assertion

The assertion decides the test, and above all the direction of the test.

Assertion	The question	Typical tests
Existence / occurrence	Did it happen, does it exist?	Vouch recorded items to support; confirmations; inventory observation
Completeness	Is everything recorded?	Trace source documents to the records; search for unrecorded liabilities; cut-off tests
Accuracy / valuation	Are the amounts right?	Recalculation; pricing tests; allowance and impairment review
Cut-off	Is it in the right period?	Test transactions either side of year end
Rights and obligations	Does the entity own it or owe it?	Contracts, title documents, loan agreements
Classification and presentation	Is it in the right account and disclosed properly?	Review of coding; disclosure checklists

Our guide to vouching vs tracing goes deeper into why direction matters.

Examples by audit area

Cash. Confirm balances with the bank, test the year-end bank reconciliation (agree reconciling items to the next period's statements), and review for unusual transfers near year end.

Accounts receivable. Send confirmations for a sample of balances. For non-replies, test subsequent receipts, and vouch the remaining items to invoices and shipping documents. Test the allowance against aging and history.

Revenue. Vouch a sample of recorded revenue to contracts, invoices and proof of delivery. Test cut-off around year end. Use analytics by month, product or location where the data is reliable.

Inventory. Observe the count, test counts both ways (floor to sheet and sheet to floor), test pricing to invoices or cost records, and look for slow-moving items.

Accounts payable and expenses. Vouch expense samples to invoices and approvals for occurrence. Run the search for unrecorded liabilities for completeness. Analytics work well for recurring costs.

Payroll. Substantive analytics are usually strong here, with tests of details for new hires, terminations and unusual payments.

How many items to test

For tests of details you usually combine:

• Key items: everything above a threshold, or items that are unusual. These are tested 100%.
• A representative sample of the rest, sized from materiality, the risk of material misstatement, and how much assurance you already have from controls and analytics.

Monetary unit sampling suits overstatement testing because larger items are more likely to be selected. For completeness, the sample has to come from a population outside the records.

When you find a misstatement in a sample, project it to the population. Do not just correct the item you found.

Timing

You can perform substantive procedures at an interim date, but then you need procedures covering the period from interim to year end: usually a roll-forward of the balance plus analytics or more testing. Cut-off and year-end balances are almost always tested at or after year end.

Documenting the work

Each test needs a clear objective, the population and how the sample was chosen, what was done to each item, the result, and a conclusion that ties back to the assertion. Our guide to audit working papers covers what reviewers and inspectors look for.

Where Warp fits

The most time-consuming substantive tests are the document-heavy tests of details: vouching samples to invoices, receipts and contracts, and journal entry testing. Warp runs those and returns workpapers where every conclusion links to the source page with the reasoning shown, so your team reviews and signs off. See test of details and journal entry testing in Warp.