Glossary

Variance analysis

Variance analysis in accounting is the process of comparing an actual financial result with a named baseline, such as a budget, forecast, or prior period, then investigating and documenting why the difference occurred. It measures the gap, separates supported drivers, and determines whether follow-up is needed.

Why variance analysis matters

A total difference shows what changed, but it does not prove why. Finance teams use variance analysis to connect a comparable result with evidence, distinguish operating causes from timing or scope effects, and decide whether an explanation, control action, forecast update, or accounting review is appropriate.

The choice of baseline determines what a result can tell you. Comparing with an approved budget tests performance against that plan; comparing with a later forecast tests performance against revised expectations. Keep the version identifiable. Replacing the original baseline after seeing the result can obscure the question the analysis was intended to answer.

Activity matters as well. If output changes, some variable costs may move with it while fixed costs behave differently. A comparison adjusted for actual activity, often called a flexible budget, can help separate the effect of volume from spending performance. State what was adjusted and why, so a reader can distinguish a volume effect from a change in the cost of producing the same output.

What is a variance in accounting?

A variance in accounting is the difference between an actual amount and the amount it is compared with, such as a budget, forecast, standard cost, or prior period. A variance is favorable when it improves the result, such as revenue above budget or cost below budget, and unfavorable when it worsens the result. The sign alone does not decide which: a positive variance is favorable for revenue and unfavorable for an expense, so every report should state its convention.

Variance analysis formula

The basic formula is variance = actual result – baseline. The percentage variance is (actual – baseline) / baseline × 100, which requires a positive, nonzero baseline. For example, a budget of $50,000 and an actual of $60,000 give a variance of $10,000, or 20% of the budget. A percentage helps compare lines of different size, while the amount shows what the difference is worth.

How does variance analysis work?

  1. Define a comparable basis. Name the entity, period, currency, account or measure, and baseline. Adjust the baseline when activity levels must be comparable.
  2. Calculate the variance. State the convention. Using actual minus baseline, the amount variance equals actual less baseline, and the percentage equals that difference divided by a positive, nonzero baseline, multiplied by 100.
  3. Investigate supported drivers. Use invoices, payroll records, transaction detail, approved operating changes, and other evidence to explain causes without counting the same driver twice.
  4. Document and review the response. Record the explanation, evidence, owner, and required action. A threshold breach prompts investigation, but it does not automatically require a journal entry.

Variance analysis example

A US business budgeted $1,200,000 of operating expenses for Q2 and recorded $1,260,000 of actual expense. Using actual minus budget, the variance is $60,000, or 5%, unfavorable for a cost. The evidence-backed cause bridge is $25,000 of overtime, $18,000 for additional software licenses used during the quarter, $12,000 of additional site travel, $8,000 from a utility rate increase, and $3,000 of lower supply use. The drivers reconcile exactly: $25,000 + $18,000 + $12,000 + $8,000 – $3,000 = $60,000. The calculation confirms the bridge, while payroll, invoices, approvals, utility bills, and purchasing records support the causes.

A useful next step is to identify which drivers are expected to recur. Additional licenses may continue if they support an ongoing need, while particular travel may relate to a completed assignment. Those are questions for the responsible teams, not conclusions supplied by the arithmetic. Their answers determine whether the finding informs future expectations, requires a control response, or simply explains the quarter.

The bridge also needs a completeness check: each amount should be supported, allocated to a distinct driver, and included only once. If the available evidence explains only part of the total, show the remainder as unexplained. Assigning that remainder to a convenient cause would make the report appear complete while weakening its decision value.

Types of variance analysis

Management accounting usually groups variances into four families: sales, materials, labor, and overhead. Each family splits a total difference into a price or rate effect and a quantity or volume effect, so the team can see whether it paid more, used more, or sold a different amount. The formulas below use actual minus budget or standard; for costs a positive result is unfavorable, and for sales it is favorable.

VarianceFormulaWhat it isolates
Sales price(Actual price – budget price) × actual unitsSelling at a different price
Sales volume(Actual units – budget units) × budget priceSelling a different quantity
Material price(Actual price – standard price) × actual quantityPaying a different price for inputs
Material quantity(Actual quantity – standard quantity) × standard priceUsing more or less input than the standard
Labor rate(Actual rate – standard rate) × actual hoursPaying a different hourly rate
Labor efficiency(Actual hours – standard hours) × standard rateWorking more or fewer hours than the standard
Overhead spendingActual overhead – budgeted overhead for the actual activitySpending on indirect costs

A materials example shows how the split works. The standard is 2 kg of input per unit at $5.00 per kg, and the business produced 1,000 units, so the standard quantity is 2,000 kg and the standard cost is $10,000. It actually used 2,100 kg bought at $4.80, an actual cost of $10,080. The price variance is ($4.80 – $5.00) × 2,100 = -$420, favorable. The quantity variance is (2,100 – 2,000) × $5.00 = $500, unfavorable. Together they explain the total: $500 – $420 = $80 unfavorable, which equals $10,080 – $10,000.

Variance analysis vs. flux analysis

Variance analysis can compare actual results with a budget, forecast, standard, or prior period. Flux analysis, also called fluctuation analysis, often emphasizes period-to-period balance changes, and it is commonly reviewed during the month-end close next to each balance sheet reconciliation. The terms overlap in practice, so each analysis should name its comparison basis. Reconciliation answers whether records agree; variance analysis asks why a comparable result changed.

For either comparison, define investigation thresholds before interpreting the result and retain qualitative context. A large percentage on a small base can mean something different from a modest percentage on a substantial balance.

What a variance report should include

A variance report lists, for each line in scope, the baseline, the actual amount, the variance in amount and percentage, and whether it is favorable or unfavorable. Lines above the agreed threshold also carry the supported explanation, the evidence used, the owner, and the next action, and any unexplained remainder is shown as such. Reporting only the lines that breach a threshold is called management by exception: it focuses review time, but the threshold has to be set before the results are known.

How Simetrik supports variance analysis

Simetrik, an AI-native financial control platform, supports the data and control side of variance analysis rather than the budget itself. In the Operation Center dashboards and anomaly detection, teams build datasets from reconciled transactions and sources, define the KPIs to follow, and set statistical thresholds that flag unusual changes in volumes, reconciliation rates, or data quality, part of its unified dashboards and alerts. Because the comparison starts from reconciled records, a variance can be traced back to the transactions behind it.

For balances, Simetrik’s account reconciliation and close controls compare the ERP balance with the reconciled balance account by account and period by period. Each difference receives a structured justification with its cause, amount, owner, and resolution date, and a preparer and a certifier review it. For broader financial control and visibility for finance teams, the same evidence supports the analysis, while the finance team chooses the baseline, interprets causes, and approves actions.

Frequently asked questions

How do you calculate a variance percentage?

With the actual-minus-baseline convention and a positive, nonzero baseline, divide the amount variance by the baseline and multiply by 100. A zero baseline has no meaningful percentage, and a negative baseline requires an explicitly defined convention.

What does a 20% variance mean?

A 20% variance means the actual result differs from the baseline by one fifth of the baseline. A $60,000 actual against a $50,000 budget is a 20% variance; whether it is favorable depends on whether the line is revenue or cost.

Is a favorable variance always good?

No. Favorable or adverse depends on economic meaning, not the sign alone. Lower cost may be favorable, but it can also reflect delayed invoices, lower activity, or work that was not completed.

What are the four main types of variance analysis?

The four families most textbooks use are sales, materials, labor, and overhead variances. Each one separates a price or rate effect from a quantity, volume, or efficiency effect.

Is variance analysis the same as reconciliation?

No. A reconciliation checks whether two records of the same item agree, such as the ledger and the bank statement. Variance analysis compares a result with a baseline and explains why it differs. A reconciled balance is a better starting point for variance analysis, because the analysis then explains real changes rather than recording errors.

Can AI explain a variance?

Yes, AI can point to where a variance comes from, but the cause still needs evidence and a person’s conclusion. In Simetrik, AI Copilot Datasets can build the comparison dataset from a natural-language description, and anomaly detection groups related signals into an incident and proposes a probable root cause. The underlying matching runs on deterministic rules, and the finance team confirms the explanation.

What is needed to configure variance review in Simetrik?

A configured variance control needs supported balance and evidence data, account and period mappings, a comparison basis, thresholds, and review responsibilities defined by the finance team. Simetrik organizes the records and surfaces differences for review, but it does not approve a cause or corrective action autonomously.

Ready to transform your reconciliation workflows?

This site is registered on wpml.org as a development site. Switch to a production site key to remove this banner.