15.1 Ratios, Variances, Trends, and Benchmarking

Analytical Review as an Engagement Procedure

Internal auditors use analytical review to decide whether recorded financial results, operating statistics, and other information are consistent with what should have occurred given the activity under review. On the 2025 CIA Part 2 syllabus this is Section B5a: analyze ratios, variances, trends, financial and nonfinancial information, and benchmarking results. The skill sits under the Global Internal Audit Standards Principle 14 (Conduct Engagement Work) and Standard 14.2 (Analyses and Potential Engagement Findings). You are not ranking stocks, building a discounted cash-flow model, or picking a peer multiple for a valuation. You are answering an engagement question: given this activity's objectives, volumes, rates, mix, seasonality, and the evaluation criteria locked in planning, is this result plausible?

That purpose change is the first exam filter. A CFA-style current ratio, debt-to-equity, or EBIT margin can appear in an engagement if the objective is solvency, capital structure, or earnings quality for the activity. For a warehouse, claims unit, or plant, the more typical tools are operational ratios: lines picked per labor hour, freight cost per shipment, shrink as a percent of throughput, cycle-count accuracy, claims closed per FTE, first-pass yield, purchase-order cycle time, and the percentage of invoices that completed three-way match. The ratio is a lens. The engagement objective chooses the lens.

Form a Plausible Expectation First

The most reliable CIA Part 2 trap in this chapter is reverse-engineering an "expectation" after you have already seen the actual number. That is not analysis. That is confirmation bias. Standard 14.2 requires analyses that can support potential findings. An analysis that starts with the actual and then hunts for a story that makes the actual look reasonable cannot do that work.

Before you treat actual results as the baseline, build a plausible independent expectation from information you can defend in the workpapers:

• Prior-period results adjusted for known changes—volume, wage or commodity rates, product mix, one-time events, new facilities, system cutovers, strikes, recalls, or a 3PL switch
• The approved budget or forecast, only after you understand the assumptions inside it (fixed versus flexible, stretch versus operational, whether volume was locked)
• Independent operational data that does not come from the same recording process you are testing—bills of lading, production counts, badge-swipe hours, meter readings, claim inventories, shipment files
• Relevant internal or external benchmarks that are actually comparable on definition, technology, service promise, and accounting treatment

Plausible does not mean precise to the dollar. It means a competent auditor could explain why this range is what should have happened. If last year overtime was $800,000 and this year the hub processed 20 percent more lines, wage rates rose 4 percent, and a night shift started in March, "same as last year" is not a plausible expectation. A defensible expectation starts with $800,000 × 1.20 × 1.04 and then adds a night-shift premium you can source. You then set a threshold for investigation—percent, dollars, or both—tied to engagement materiality and to how precise the expectation can be. Only then do you compare recorded amounts. Differences inside the threshold support the recorded amount for this procedure. Differences outside the threshold are unexpected and must be investigated.

Document the expectation, the sources, the threshold, and the comparison. If the workpaper only shows "actual exceeded budget by 12 percent—investigate," you skipped the procedure the exam is testing.

Budget Versus Actual, Then Disaggregate

Budget-versus-actual is the variance analysis you will see most often. Treat the budget as a criterion only when it is a relevant, reliable expected state for the objective. A stretch target that operations never used to run the dock is a poor criterion for "did spending stay in control." A volume-flexible budget that supervisors actually used is a strong one.

Never stop at the total-dollar box. A $48,000 (12 percent) freight overrun on a $400,000 budget looks like a finding until you learn shipments rose 15 percent and the fuel surcharge matched the budget file. Cost per shipment may be favorable. The condition is not "over budget." The condition, if any, is whatever remains after you compare actual to a volume-adjusted expectation.

Year-over-year trend analysis asks a different question: what is the direction, slope, and break across periods? Three years of rising shrink, a jump after a warehouse-management-system go-live, or flat headcount while volume doubled each tell a different story. Known changes must enter the expectation. If you cannot explain a break with corroborated facts, you have an unexpected trend—not yet a finding, but a difference Standard 14.2 requires you to analyze.

Benchmarks, Financial and Nonfinancial Pairing, and a Worked Freight Example

Benchmarking compares the activity to a relevant peer: another warehouse in the same network, a sister plant, a contract service level, or an industry quartile. Comparability is the control. Different pick technology, different SKU cube, a tighter service promise, or freight capitalized in one site and expensed in another will manufacture a fake gap. Internal benchmarks (North Hub versus South Hub on the same system and the same network rules) are usually more reliable than a purchased industry average. External benchmarks still help as a reasonableness range. They become evaluation criteria only when planning actually adopted them (Chapter 3), not because a slide deck once showed a quartile.

Financial amounts without the operating driver are easy to game and easy to misread. Pair them.

If freight dollars rose 12 percent and shipments rose 15 percent, the combined ratio improved. If overtime dollars rose 18 percent while lines picked were flat, the operational ratio deteriorated—and that unexpected relationship is the signal.

Walk the freight numbers. Planning criterion: freight should track the flexible budget. Independent data: 15 percent more shipments; fuel surcharge rates match the budget file. Expectation: $400,000 × 1.15 = $460,000. Actual: $448,000. Difference: $12,000 favorable to the volume-adjusted expectation (about 2.6 percent). If the threshold was 5 percent or $25,000, this variance is not unexpected. The raw budget-versus-actual of plus 12 percent was a red herring. If actual had been $520,000 against the $460,000 expectation, the $60,000 (13 percent) gap exceeds the threshold. Next step is investigation—mode mix, detention, misposted accounts, a carrier change—not an automatic finding and not an automatic fraud conclusion.

Analytics identify unexpected relationships. They do not, by themselves, prove a condition. The condition in a finding must still be evidenced (Chapter 12; Standards 14.1 and 14.2). A ratio that sits on an industry average also does not prove the control operated. When a stem gives budget, last year, and a known volume change, compute the expectation before you react to the actual. If the question asks what to do first, form the expectation and the threshold. If it asks whether to report a finding from a raw budget overrun, disaggregate volume, rate, and mix. If it offers a solvency ratio for an efficiency objective, take the operational ratio instead.