Analytical procedures are not just year-on-year variance checks added to an audit file at the end. Under ICAI's SA 520, they are evaluations of financial information through plausible relationships among financial and non-financial data, together with investigation of significant or inconsistent fluctuations. Used properly, they can help an auditor obtain substantive evidence and also challenge whether the financial statements make sense as a whole.
Where SA 520 fits in an audit
SA 520 specifically deals with substantive analytical procedures and analytical procedures performed near the end of the audit. Analytical procedures used as risk-assessment procedures are addressed by SA 315. ICAI's official Engagement and Quality Control Standards library lists SA 520 within the audit-evidence standards, while the official SA 520 text sets out the detailed requirements.
The distinction matters. A planning-stage comparison may identify an unusual margin and therefore a risk. A substantive analytical procedure is designed to obtain audit evidence about an assertion. A final analytical review asks whether the financial statements, taken together, remain consistent with the auditor's understanding of the entity.
What makes an analytical procedure substantive?
A substantive analytical procedure needs more than a percentage movement and a management explanation. SA 520 requires the auditor to consider four linked questions: whether the procedure is suitable for the assertion, whether the underlying data is reliable, whether the auditor can develop a sufficiently precise expectation, and what difference from that expectation can be accepted without further investigation.
1. Start with the assertion and risk
Analytical procedures are generally more useful for large volumes of transactions that behave predictably over time. Payroll for a stable workforce, rental income from known occupied units, interest expense on known borrowings, and certain recurring sales-margin relationships may lend themselves to prediction. A highly judgmental, unusual or non-routine balance may require greater reliance on tests of details.
2. Test whether the data is reliable
A sophisticated model built on unreliable inputs is weak audit evidence. SA 520 says reliability is influenced by the source, comparability, nature and relevance of the information and controls over its preparation. External information may be stronger than internally generated data in appropriate circumstances, while management budgets need scrutiny over how they were prepared and whether they represent realistic expectations rather than aspirational targets.
Non-financial data can be particularly useful. Payroll expense can be related to employee numbers and salary rates; rental income can be related to units, contracted rent and occupancy; production quantities can be compared with raw-material consumption. But the auditor should establish that the non-financial data itself is complete and accurate enough for the purpose.
3. Build an expectation precise enough to detect a material problem
The expectation should be capable of identifying a misstatement that could matter to the financial statements. Broad comparisons can hide offsetting movements. Disaggregating revenue by product, geography, month or business unit can often produce a more precise expectation than comparing only total annual revenue.
Precision also depends on predictability. Gross margins may be reasonably stable in some businesses but volatile in others because of commodity prices, product mix, discounting or currency movements. The auditor should understand the drivers before deciding that a variance is unusual.
4. Set an investigation threshold before seeing the result
SA 520 requires the auditor to determine the amount of difference between recorded and expected values that can be accepted without further investigation. That threshold is influenced by materiality and the desired level of assurance. As assessed risk increases, the acceptable unexplained difference generally becomes smaller because more persuasive evidence is required.
Worked example: payroll reasonableness test
Assume a company has 200 employees throughout the year, salary structures are stable, payroll changes are centrally approved, and reliable employee-count and salary-rate data is available. Instead of merely comparing this year's payroll with last year's figure, the auditor can construct an independent expectation from employee numbers, contractual salary rates, known increments, joining and exit dates, bonuses and statutory employer costs where relevant.
Suppose the auditor's expectation is materially below the recorded payroll. The difference should not simply be labelled a variance and cleared through inquiry. The auditor would investigate whether the gap comes from bonuses, overtime, new hires, one-time payments, incorrect source data or possible accounting errors, and obtain evidence supporting the explanation. If management's explanation and evidence are inadequate, SA 520 requires other audit procedures as necessary.
Analytical procedures versus tests of details
Neither method is automatically superior. Tests of details examine individual transactions, balances or documents. Substantive analytical procedures test plausible relationships at a broader level. SA 520 recognises that substantive procedures may consist of analytical procedures, tests of details, or a combination of both.
A practical choice depends on the assertion and the predictability of the data. For recurring rental income with verified occupancy and contractual rates, a strong predictive model may provide persuasive evidence. For unusual related-party transactions or a balance driven by complex individual contracts, detailed examination may be more responsive to risk.
How to investigate an unexpected variance
- Recheck the model and inputs. Confirm that formulas, periods, classifications and source data are correct.
- Ask for a specific explanation. “Sales increased” is not enough; identify the product, customer, price, volume or timing driver.
- Corroborate management's response. Use contracts, operational reports, board papers, invoices, production data or other appropriate evidence.
- Consider whether the variance reveals a new risk. An unexpected relationship may affect other audit areas or require a revised risk assessment.
- Perform additional procedures when needed. If the explanation is unavailable or unsupported, extend testing rather than forcing the variance to fit the original expectation.
Final analytical review is a separate discipline
SA 520 requires analytical procedures near the end of the audit to assist the auditor in forming an overall conclusion about whether the financial statements are consistent with the auditor's understanding of the entity. This is not merely a repeat of planning analytics. It is a final reasonableness challenge after audit adjustments and detailed work have been considered.
A final review can expose a relationship that individual audit sections did not make obvious: margins that no longer align with the business explanation, a working-capital movement inconsistent with reported growth, or an expense relationship that changed after late adjustments. If the review reveals a previously unrecognised risk, the auditor may need to revisit the risk assessment and planned procedures.
Documentation checklist
- State the assertion and audit objective.
- Explain why the analytical procedure is suitable for that assertion.
- Identify the financial and non-financial data used and how reliability was assessed.
- Document the expectation and the logic behind it.
- Record the acceptable difference or investigation threshold.
- Explain significant variances and retain corroborating evidence.
- Document additional procedures where explanations were insufficient.
- Record the conclusion and how it supports the audit objective.
Common mistakes
- Using only prior-year percentages without understanding current business drivers.
- Accepting management explanations without corroborating evidence.
- Using budgets without assessing how reliably they were prepared.
- Building expectations at such an aggregated level that material problems can be hidden.
- Choosing the investigation threshold after seeing the variance.
- Treating final analytical review as a checklist sign-off rather than a fresh overall reasonableness assessment.
Practical takeaway
A strong SA 520 procedure is a testable prediction, not a decorative ratio. Start with the assertion, use reliable data, build a sufficiently precise expectation, decide in advance what difference requires investigation, and corroborate explanations with evidence. When the relationship is not predictable enough, or the risk is too specific, use tests of details or combine both approaches. The value of analytical procedures comes from disciplined expectation and investigation, not from the number of ratios in the working paper.
