Substantive Analytical Procedures in Auditing: Techniques and Steps

Substantive analytical procedures in auditing are tests that evaluate financial data by building an independent expectation for an account balance and comparing that expectation to what the company recorded. They are one of the two main forms of substantive testing, alongside tests of details, and auditors choose between them based on which approach provides stronger evidence for the assertion being tested. The core idea is straightforward: if the relationships among revenue, cost, headcount, debt, interest rates, and similar variables behave predictably, an auditor can use those relationships to check whether a recorded amount is reasonable.

Analytical procedures also appear during audit planning and during the overall review at the end of an engagement, but those uses serve different purposes and are governed by different standards. This article covers the substantive stage, where the procedure functions as actual audit evidence under PCAOB AS 2305.¹

The Four Techniques Auditors Use

Trend analysis examines how an account balance has moved across multiple periods. The underlying assumption is that past patterns continue unless something specific in the business changed. If revenue jumped 40 percent while headcount and production volume stayed flat, that gap demands an explanation. It is the simplest technique and works well for income statement accounts where year-over-year comparisons are naturally meaningful.

Ratio analysis compares relationships between different financial statement items, or between financial and non-financial data. The current ratio, gross margin percentage, inventory turnover, and days sales outstanding are staples. A single ratio in isolation rarely tells you much; the value comes from comparing against prior periods or industry averages.

Reasonableness testing uses known data to build an independent estimate of what an account balance should be. An auditor estimating interest expense might multiply the average outstanding debt balance by the weighted average interest rate. If the calculated figure lands close to the recorded expense, that is strong evidence the balance is fairly stated. This method works best when the inputs are reliable and the relationship between them is straightforward.

Regression analysis is the most statistically rigorous option. It models the relationship between a dependent variable and one or more independent variables to predict an account balance. Payroll expense, for instance, can be modeled against employee headcount, average salary, and overtime hours. Because regression quantifies how much each input should influence the output, it can detect misstatements that simpler methods would miss. The trade-off is that it demands more data and statistical expertise to execute properly.

What Makes an Expectation Precise Enough

Not all analytical procedures carry the same weight as audit evidence. An expectation needs to be precise enough that a difference representing a potential material misstatement would actually get flagged for investigation.¹

Four factors drive that precision.

Disaggregation. Testing at a more detailed level increases precision. Analyzing total revenue across an entire company might mask offsetting errors that would surface if you broke the analysis down by product line or geographic segment. The more granular the data, the harder it is for misstatements to hide.

Data reliability. The precision of any expectation depends on the quality of inputs. Data from independent external sources carries more weight than internally generated figures, and data that has already been audited is more reliable than unaudited data.

Predictability. Some relationships are inherently more stable than others. Income statement accounts representing transactions over a full period tend to be more predictable than balance sheet accounts reflecting a single point in time. Accounts subject to management discretion are less predictable still.¹

Type of procedure. Regression analysis generally produces a more precise expectation than simple trend analysis, because it controls for multiple variables simultaneously. The auditor picks the technique that matches the desired level of assurance.

When a higher level of assurance is needed from analytical procedures, the auditor must use more predictable relationships and more reliable data to build the expectation. A procedure serving as the primary substantive test for a significant account needs substantially more rigor than one used to supplement other testing.

Running the Procedure Step by Step

Gathering and Verifying the Data

The foundation of any substantive analytical procedure is the data used to develop the expectation. Auditors draw on prior-period financial statements, internal budgets and forecasts, industry benchmarks, and external economic indicators. Information from independent sources outside the company is generally considered more reliable than records prepared by company staff. Within the company, data prepared by people independent of the group responsible for the account being audited is more trustworthy than data from that group itself. Information developed under a system with strong controls, and data tested in a current or prior-year audit, rank higher on the reliability scale.¹

Before relying on the results, the auditor must verify the completeness and accuracy of the underlying data. This happens one of two ways: by testing the design and operating effectiveness of controls over the financial information being used, or by performing separate procedures to confirm the data is complete and accurate.¹ Skip this step and the entire procedure produces unreliable results, no matter how sophisticated the statistical model on top.

Developing the Expectation

Once the data is assembled and verified, the auditor calculates a specific numerical expectation for the account balance under review. This figure represents what the balance should be if the underlying financial relationships are operating as expected. The expectation cannot be developed using the company’s own recorded amount or information derived from that amount, because doing so defeats the purpose of an independent check.¹

Setting the Threshold and Comparing

Next comes the threshold: the maximum acceptable difference between the expectation and the recorded amount. For public company audits under PCAOB standards, this threshold must be set at or below tolerable misstatement for the account.²

Quantitative rules of thumb have developed around materiality. The SEC has noted that some auditors treat a 5 percent threshold as a rough dividing line for materiality, while other guidance references ranges of 1 to 10 percent depending on the account and the risk level. The SEC has also warned that relying exclusively on any single percentage threshold has no basis in the accounting literature or the law.³

With the threshold set, the auditor compares the recorded amount to the expectation. If the difference falls within the threshold, the procedure provides evidence that the account is fairly stated. If it exceeds the threshold, the auditor has a significant unexpected difference that requires investigation.

Investigating Unexpected Differences

When the difference exceeds the threshold, the investigation typically starts with asking management for an explanation. Management might point to a specific business event: a large new contract, a change in pricing strategy, or an asset impairment. The auditor cannot stop there. Responses must be corroborated with independent evidence, which might mean reviewing signed contracts, examining third-party invoices, checking bank confirmations, or testing journal entries that correspond to the claimed event. If the explanation seems implausible, inconsistent with other audit evidence, or lacks sufficient detail, the auditor must perform additional procedures to resolve the matter.⁴

If explanation and corroborating evidence remain unsatisfactory, the auditor must shift to more detailed substantive testing by examining individual transactions and balances within the account. This is the point where analytical procedures hand off to tests of details.¹

Documentation the Workpapers Must Show

When an analytical procedure serves as the principal substantive test of a significant financial statement assertion, the auditor must document the expectation and the factors considered in developing it, the results of comparing the expectation to the recorded amount, and any additional procedures performed in response to significant unexpected differences along with their results.¹

Good documentation lets a reviewer follow the logic chain from start to finish: what data went into the expectation, what formulas or models were used, what the threshold was and why it was set at that level, and what the resulting variance was. Workpapers that say “performed analytical procedures over revenue, no exceptions noted” provide no evidence that real work was done, and this is an area where PCAOB inspection findings consistently identify deficiencies.

When Analytical Procedures Are Not Enough

Substantive analytical procedures have real blind spots. For significant risks of material misstatement, audit evidence from analytical procedures alone is unlikely to be sufficient. Those high-risk areas almost always require tests of details, either on their own or combined with analytical procedures.¹

Analytical procedures are also poorly suited to detecting fraud. Management override of controls can produce artificial changes to the very financial relationships the auditor is analyzing. If someone manipulated the numbers, the manipulated relationships might look perfectly normal to the analytical test because the fraud was designed to maintain expected patterns.¹

Several other conditions reduce effectiveness. Relationships involving only balance sheet items are inherently less predictable because they represent a snapshot at a single point in time rather than cumulative activity over a period. Accounts driven by management judgment, such as when to incur maintenance expense or when to ramp up advertising, produce less predictable patterns. Companies undergoing rapid change, entering new markets, or operating in volatile industries generate relationships that shift from period to period, making historical patterns unreliable as benchmarks. And when the data needed to test a specific assertion is not readily available, testing the underlying transaction records directly is often more effective. Completeness of sales, for example, may be better tested by tracing shipping records than by running a ratio analysis.

Public Company vs. Private Company Requirements

Public company audits fall under PCAOB standards, primarily AS 2305, while private company audits follow AICPA standards under AU-C Section 520. The core concepts overlap, but several requirements are stricter on the public company side.

AS 2305 explicitly prohibits using the company’s own recorded amount to develop the auditor’s expectation. AU-C 520 does not include this prohibition.⁵

AS 2305 requires the threshold to be set at or below tolerable misstatement and demands that the auditor address the risk that the difference represents a material misstatement. AU-C 520 does not specify these threshold requirements.⁵

AS 2305 also states that determining whether a relationship is sufficiently plausible and predictable must extend beyond inquiry. AU-C 520 has no equivalent requirement.⁵

Both standards require investigation of significant unexpected differences, but AS 2305 requires that procedures extend beyond inquiry of management. AU-C 520 requires the auditor to inquire of management and perform other procedures “as necessary in the circumstances,” giving more room for judgment.⁵

The practical effect is that public company audits demand more structure and more documentation around analytical procedures. If you are moving between public and private company work, the differences in threshold-setting and investigation requirements are the areas most likely to trip you up.

Biases That Undermine the Work

Even when the mechanics are executed correctly, the human side of analytical procedures creates risk. Two biases show up repeatedly in audit quality research.

Confirmation bias leads auditors who start with the assumption that an account is fairly stated to interpret ambiguous evidence in favor of that assumption. Signals that something is wrong get discounted. This is particularly dangerous during the investigation phase: if you already believe the numbers are fine, a plausible-sounding explanation from the CFO feels like enough.

Availability bias causes auditors to anchor on the most easily recalled explanation for a variance rather than considering the full range of possibilities. When management provides a quick, confident answer for an unexpected fluctuation, that explanation can crowd out alternative hypotheses. The fix is deliberate: before accepting any explanation, force yourself to generate at least one competing explanation and evaluate both against the evidence.

These biases do not appear in the standards as formal requirements, but they explain a meaningful share of the deficiencies regulators identify during inspections. Treat every management explanation as a hypothesis to be tested rather than an answer to be confirmed.

1
Public Company Accounting Oversight Board. AS 2305: Substantive Analytical Procedures
2
Public Company Accounting Oversight Board. Statement In Support of Substantive Analytical Procedures
3
U.S. Securities and Exchange Commission. Staff Accounting Bulletin No. 99 – Materiality
4
Public Company Accounting Oversight Board. AS 2810: Evaluating Audit Results
5
Public Company Accounting Oversight Board. Comparison of Proposed AS 2305 with ISA 520 and AU-C Section 520