Think of analytical procedures as the auditor's version of a sanity check — before diving into detailed vouching, you step back and ask: does this data even make sense? That's the heart of SA 520.
SA 520 requires analytical procedures at two mandatory stages and permits them at a third. First, during risk assessment (planning stage) — here, you use them to understand the client's business and spot areas that need more attention. Second, at the overall review stage (near completion) — you check that the final financial statements are consistent with your understanding of the entity. These two are compulsory. Third, analytical procedures can optionally be used as substantive procedures (instead of, or alongside, detailed tests of transactions) — but only when they're more efficient and effective.
The standard identifies several types of analytical procedures: ratio analysis (gross profit ratio, current ratio), trend analysis (comparing this year's figures to last year's), reasonableness tests (estimating expected interest expense based on average loan balance × rate), and regression analysis (more advanced, rarely tested at Inter level). The key concept is expectation — you form an expectation of what a figure should be, compare it to what it is, and investigate any significant differences beyond a pre-set threshold of acceptable difference.
When using analytical procedures as substantive procedures, SA 520 says four things must be considered: (1) the suitability of the procedure for the assertion being tested, (2) the reliability of the data used, (3) whether the expectation is precise enough to detect a material misstatement, and (4) the acceptable amount of difference without investigation. If you find an unusual fluctuation and management's explanation is plausible, you still need corroborating audit evidence — you can't just accept what the client says. This is a favourite 4-mark exam question — 'What should an auditor do when analytical procedures reveal significant unexpected fluctuations?' Answer: obtain explanations, corroborate them with evidence, and if not satisfactorily explained, consider revision of risk assessment.