How to Perform Analytical Procedures Under ISA 520

Key Takeaways

ISA 520.5 requires four steps: determine suitability, evaluate data reliability, develop an independent expectation precise enough to detect material misstatement, and set a quantified threshold for acceptable difference.
A comparison is not an expectation. Comparing this year’s revenue to last year’s without adjusting for known changes does not satisfy ISA 520.A5.
The threshold must be stated as an absolute currency amount tied to performance materiality, not just a percentage variance. A 2% variance on a €50M line can exceed materiality by a factor of 2.5.
ISA 520.7 requires both inquiry of management and corroborating evidence for every variance that exceeds the threshold. Recording only the verbal explanation is the single most common ISA 520 inspection finding.

You’re reviewing a trial balance where revenue increased 14% year-on-year. The client says it’s organic growth. Your senior says it looks fine. But ISA 520.5 doesn’t ask whether revenue “looks fine.” It asks whether you developed an independent expectation, set a threshold for what counts as a significant difference, and investigated every variance that exceeded it. Most files skip at least one of those steps.

To perform analytical procedures under ISA 520, develop an independent expectation for each balance or class of transactions using reliable data, set a quantified threshold based on performance materiality, compare the expectation to the recorded amount, and investigate any difference that exceeds the threshold with corroborating evidence documented in the working paper.

What ISA 520.5 actually requires (and what most files miss)

ISA 520.5 sets out four requirements for a substantive analytical procedure. You need to determine the suitability of the procedure for the assertion you’re testing. You need to evaluate the reliability of the data you’re using to develop the expectation. You need to develop an expectation of the recorded amount that is precise enough to identify a misstatement (individually or in aggregate) at the level you’ve set as material. And you need to determine the amount of difference from the expected value that’s acceptable without further investigation.

Most files satisfy the first requirement by default (they pick a procedure) and partially satisfy the second (they mention the data source). The gap sits in requirements three and four. A file that compares this year’s revenue to last year’s revenue and notes “no significant variance” hasn’t developed an expectation. It’s made a comparison. ISA 520.A5 draws this distinction explicitly: an expectation incorporates relevant factors that would affect the balance, while a simple comparison does not.

The distinction matters because a comparison alone doesn’t produce audit evidence at the assertion level. ISA 520.5(c) requires precision at the assertion level. If your expectation isn’t precise enough to detect a material misstatement within that specific assertion, the procedure doesn’t satisfy the standard, even if the numbers happen to match.

How to develop an independent expectation

ISA 520.A12 identifies what makes an expectation reliable. The expectation should use data from a source independent of the general ledger where possible. It should incorporate factors that would change the expected outcome (price changes, volume shifts, new contracts, discontinued lines). And it should be developed at a level of disaggregation that makes it precise enough to detect misstatement at performance materiality.

Start with external or non-financial data. If you’re testing revenue for a property management firm, the number of managed units multiplied by the average monthly fee is more reliable than last year’s revenue plus a growth assumption. Industry data, contracts, physical capacity, headcount, pricing schedules, and production records all sit outside the general ledger and give you genuine independence.

The type of data you use directly affects the precision of your expectation. ISA 520.A5 distinguishes between expectations developed from independent sources and those built on the client’s own prior-year data. An expectation built from a signed contract register, an external price index, or a production log has inherent reliability because it originates outside the financial reporting process. An expectation built from last year’s audited figures adjusted by a growth factor the client provided is weaker. It isn’t prohibited, but ISA 520.A13 requires you to evaluate whether the adjustment is itself reliable, and your working paper needs to explain why.

When external data isn’t available, use prior-year audited figures adjusted for known changes. ISA 520.A13 permits this, but your expectation must document the adjustments. “Revenue grew because the client said so” doesn’t satisfy the standard. “Revenue grew €1.2M because the client added 14 new managed properties at an average monthly fee of €7,100, confirmed against the signed management agreements” does.

Avoid using the client’s budget as your expectation

ISA 520.A16 explicitly warns about this: if the client’s budget was prepared with the same data and assumptions that produced the financial statements, comparing the two gives you almost no independent evidence. The expectation must be built on data that could tell you a different story from the one the financial statements are telling.

Disaggregation is the other reliability lever. ISA 520.A11 notes that analytical procedures applied to disaggregated data produce more precise expectations. An entity-level gross margin that’s stable year-on-year can mask a five-point margin collapse in one division, offset by improvement elsewhere. If the client has segments, run the procedure by segment. If it has product lines, run it by product line. The Ciferi analytical review tool automates the comparison and variance computation once you’ve set the expectation inputs.

For entities where disaggregated data isn’t readily available, consider whether you can obtain it from operational systems rather than the general ledger. A logistics company’s transport management system will have revenue by route, by vehicle type, and by customer. A property manager’s operations system will have revenue by building and by tenant. These operational breakdowns often don’t exist in the GL at the same level of detail, and that’s precisely what makes them useful for developing an independent expectation.

Setting a threshold that connects to performance materiality

ISA 520.5(d) requires you to determine “the amount of any difference… that is acceptable without further investigation.” This is your threshold, and it needs a documented basis.

The starting point is performance materiality as set under ISA 320.11. Your threshold for the analytical procedure should not exceed performance materiality for the assertion you’re testing. In practice, most firms set it somewhere between 50% and 75% of performance materiality, depending on the inherent risk of the balance. A higher-risk area gets a tighter threshold. A lower-risk area can tolerate a wider one.

How you arrive at that percentage matters. The threshold isn’t arbitrary. ISA 520.A18 ties it to two factors: the materiality level and the desired level of assurance from the procedure. If you’re relying primarily on the analytical procedure as your substantive evidence for a balance (rather than supplementing it with detail testing), you need a higher level of assurance, which means a tighter threshold. If the analytical procedure is supplementary to a programme of detail testing, a wider threshold is defensible because the procedure is contributing less of the total evidence.

Document the rationale. A threshold set at €180,000 because performance materiality is €240,000 and the balance carries moderate inherent risk is defensible. A threshold with no stated connection to materiality is not. ISA 520.A18 confirms that the acceptable difference depends on the materiality level and the desired level of assurance from the procedure.

Use absolute amounts, not just percentages

One subtlety that files often miss: the threshold applies to the absolute difference between the expectation and the recorded amount, not to the percentage variance. A 2% variance on a €50M revenue line is €1M. If performance materiality is €400,000, that 2% exceeds it by a factor of 2.5. Percentage-only analysis without a euro-denominated threshold doesn’t satisfy ISA 520.5(d).

Different balances on the same engagement can have different thresholds. Revenue tested as a primary substantive analytical procedure (high reliance) might get a threshold at 50% of performance materiality. Administrative expenses tested as a supplementary analytical procedure (with detail testing on the largest line items) might get a threshold at 75%. Document each threshold separately, with its own rationale, in the relevant working paper section.

How to investigate variances properly

When the difference between your expectation and the recorded amount exceeds the threshold, ISA 520.7 requires investigation. The investigation has two parts: inquire of management and obtain corroborating evidence.

ISA 520.7(a) requires you to inquire of management and obtain appropriate audit evidence relevant to management’s response. This means you don’t just ask why the variance exists. You ask, then you test whether the explanation is consistent with other evidence on the file. Management says the €320,000 revenue variance is due to a large contract signed in Q4. You obtain the signed contract, check that invoicing matches the contract terms, and confirm that the revenue recognition aligns with IFRS 15.31 through 15.35.

The corroborating evidence must be proportional to the size of the variance relative to materiality. A variance of €50,000 against a performance materiality of €400,000 warrants a lighter touch (inquiry plus one supporting document). A variance of €380,000 against the same materiality needs multiple sources of corroboration because the unexplained amount is close to the point where it could affect your opinion.

ISA 520.7(b) adds a second layer. If management can’t provide an explanation (or the explanation doesn’t hold), you need to apply other audit procedures to determine whether the variance represents a misstatement. This might mean extending your substantive testing, performing additional detail testing on the balance, or reconsidering the assessed risk of material misstatement under ISA 315.

Record what you ruled out, not just what you concluded

If management offered two possible explanations for a variance and you investigated both, record both investigations even if only one proved correct. The reviewer sees your thought process, and ISA 230.8 requires documentation of significant judgments. An investigation that considered and eliminated alternative explanations demonstrates more professional scepticism than one that accepted the first answer.

The investigation isn’t complete until you’ve recorded the outcome in the working paper. That outcome should state what the variance was (in euros, not just percentage), what management said, what evidence you obtained, and whether the variance has been resolved or remains an unadjusted difference. Leaving a variance flagged with no documented resolution is the single most common ISA 520 finding in inspection reports.

What ISA 520.6 requires at completion

ISA 520.6 requires you to design and perform analytical procedures near the end of the audit that assist you in forming an overall conclusion on whether the financial statements are consistent with your understanding of the client. This is separate from substantive analytical procedures under ISA 520.5. The completion procedure isn’t optional, and it serves a different purpose.

The completion analytical review is a reasonableness check, not a substantive test. You’re asking whether the financial statements as a whole make sense given everything you’ve learned during the engagement. If revenue grew 14% but headcount dropped 20% and there was no price increase, that inconsistency needs investigation even if your substantive procedures on revenue passed.

ISA 520.A20 clarifies that the procedures at completion can use data aggregated at a higher level than substantive analytical procedures. Entity-level ratios, trend comparisons, and financial statement level analysis are appropriate here. The level of precision required is lower because the purpose is different: you’re looking for anomalies that might indicate a previously unrecognised risk of material misstatement, not building evidence for a specific assertion.

Document the completion analytical procedures in a dedicated section of your working paper, not mixed in with the substantive procedures. The reviewer needs to see that you performed them as a distinct step after fieldwork. Record the ratios or comparisons you performed, the expected relationships, any anomalies you identified, and how you resolved them. If no anomalies were identified, a brief note confirming this (with the specific comparisons you made) is sufficient. “Completion analytics performed, no issues noted” without specifying what you actually looked at does not satisfy ISA 520.6.

One pattern that strengthens the completion review: compare the financial statements not just to prior year but to the expectations you formed during planning. If you expected revenue growth of 8% based on signed contracts and the final figure shows 14% growth, that inconsistency warrants investigation at the completion stage even if your substantive analytical procedure on revenue passed. The completion review catches anomalies that fall between the cracks of individual account-level testing.

ISA 520.6 also functions as a safety net for the overall audit. If the completion analytical review reveals something inconsistent, it may indicate that your risk assessment was incomplete, that a significant risk was missed, or that the audit evidence obtained during fieldwork needs reconsideration. Treat the completion review as a genuine analytical exercise, not a sign-off checkbox.

The Ciferi financial ratios calculator produces the standard ratio set for completion analytics, including year-on-year comparisons and industry benchmarks where available.

Worked example: Van der Berg Holding N.V.

Client profile: Van der Berg Holding N.V. is a Dutch property management company. Revenue €28M. 340 managed commercial units across four provinces. Performance materiality €210,000 (set under ISA 320.11 at 75% of overall materiality of €280,000).

Step 1: Determine the procedure’s suitability

The engagement team selects revenue as a balance suitable for substantive analytical procedures because revenue is driven by a stable, measurable factor (number of units × average fee) and the client’s revenue model hasn’t changed.

Documentation note: Record why the procedure is suitable for the assertion (occurrence and completeness of revenue). Reference ISA 520.5(a).

Step 2: Develop the expectation

The team obtains the property management register directly from the client’s operations department (not from the finance team) showing 340 units under management at 31 December. The average monthly management fee per unit is €6,860, confirmed against a sample of 25 signed management agreements.

Expected revenue: 340 units × €6,860 × 12 months = €27,988,800.

Documentation note: Record the data source (operations register, dated), the sample of agreements checked, and the calculation. Reference ISA 520.5(b) and ISA 520.A12 on data reliability.

Step 3: Set the threshold

Revenue carries moderate inherent risk. The threshold is set at 60% of performance materiality: €210,000 × 60% = €126,000.

Documentation note: Record the threshold, the basis (percentage of performance materiality), and the risk assessment driving the percentage. Reference ISA 520.5(d).

Step 4: Compare and identify the variance

Recorded revenue: €28,410,000. Expected revenue: €27,988,800. Difference: €421,200. This exceeds the threshold of €126,000.

Documentation note: Record the comparison, the calculated difference, and flag that investigation is required under ISA 520.7.

Step 5: Investigate the variance

Management explains that 12 new units were onboarded between September and December at higher-than-average fees (commercial logistics units at €9,200 per month). The team obtains the onboarding schedule from operations, confirms 12 new units with signed agreements, and recalculates: 12 units × €9,200 × 3.5 months average = €386,400. Remaining unexplained variance: €421,200 minus €386,400 = €34,800. This is below the threshold. The variance is resolved.

Documentation note: Record management’s explanation, the corroborating evidence obtained (onboarding schedule, signed agreements for new units), the recalculation, and the conclusion that the remaining variance is below threshold. Reference ISA 520.7(a).

The reviewer sees a file where every step links back to a specific ISA paragraph, the expectation is built on data outside the general ledger, the threshold has a documented basis, and the variance is resolved with corroborating evidence.

Your ISA 520 checklist for tomorrow’s file

Confirm that the balance you’re testing is suitable for a substantive analytical procedure under ISA 520.5(a). If the balance is volatile, unpredictable, or lacks reliable underlying drivers, detail testing may be more appropriate.
Build your expectation using data from outside the general ledger. Contracts, operational records, industry benchmarks, or prior-year audited figures adjusted for known changes. Record the source and date of the data in the working paper (ISA 520.5(b)).
Disaggregate where the client has segments, product lines, or geographic divisions. Entity-level comparisons miss offsetting variances (ISA 520.A11).
Set a quantified threshold in euros (or your local currency), stated as a percentage of performance materiality, and record the risk assessment driving that percentage (ISA 520.5(d)).
Investigate every variance that exceeds the threshold. Record management’s explanation, the corroborating evidence obtained, and whether the variance is resolved or remains an unadjusted difference (ISA 520.7).
At completion, re-evaluate whether the analytical procedure gave you the evidence you planned. If it didn’t (because variances couldn’t be explained or the data turned out to be unreliable), extend your substantive testing rather than leaving a weak procedure on the file.

Common mistakes reviewers flag

Using prior-year figures as the “expectation” without documenting any adjustments for known changes. The FRC’s 2022–23 inspection cycle flagged this as a recurring finding in non-Big 4 files: a comparison is not an expectation under ISA 520.A5.
Setting a threshold as a percentage variance (e.g., “anything over 10% is investigated”) without converting it to an absolute euro amount tied to performance materiality. ISA 520.5(d) requires a quantified acceptable difference, not a rule of thumb.
Recording the variance and management’s verbal explanation but not obtaining corroborating evidence. ISA 520.7(a) requires audit evidence relevant to management’s response, not just the response itself.

Frequently asked questions

What does ISA 520.5 require for a substantive analytical procedure?

ISA 520.5 sets out four requirements: determine the suitability of the procedure for the assertion being tested, evaluate the reliability of the data used to develop the expectation, develop an expectation precise enough to identify a misstatement at the material level, and determine the amount of difference from the expected value that is acceptable without further investigation (the threshold).

How do you develop an independent expectation under ISA 520?

Use data from a source independent of the general ledger where possible: contracts, operational records, industry benchmarks, or production data. Incorporate factors that would change the expected outcome (price changes, volume shifts, new contracts). If external data is unavailable, use prior-year audited figures adjusted for known, documented changes. The expectation should be developed at a level of disaggregation precise enough to detect misstatement at performance materiality.

How should the threshold for an analytical procedure connect to performance materiality?

The threshold should not exceed performance materiality for the assertion being tested. Most firms set it between 50% and 75% of performance materiality, depending on the inherent risk of the balance. A higher-risk area gets a tighter threshold. The threshold must be stated as an absolute euro (or local currency) amount, not just a percentage variance, and the rationale connecting it to risk and materiality must be documented.

What must you do when a variance exceeds the threshold under ISA 520.7?

ISA 520.7 requires two steps: inquire of management and obtain corroborating audit evidence relevant to management’s response. The corroborating evidence must be proportional to the size of the variance relative to materiality. If management cannot provide an explanation or the explanation does not hold, apply other audit procedures to determine whether the variance represents a misstatement. Document the variance amount, management’s explanation, the evidence obtained, and whether the variance is resolved or remains an unadjusted difference.

What is the difference between substantive analytical procedures and completion analytical procedures under ISA 520?

Substantive analytical procedures under ISA 520.5 produce audit evidence for specific assertions and require precise expectations, quantified thresholds, and detailed investigation of variances. Completion analytical procedures under ISA 520.6 are performed near the end of the audit to assess whether the financial statements as a whole are consistent with the auditor’s understanding of the entity. Completion procedures can use more aggregated data and serve as a reasonableness check rather than a source of assertion-level evidence.