Variance Analysis

By Nupur Mittal

Co-founder at OCTA — works with accounting firms on month-end close and reconciliation automation.

Variance analysis is the process of comparing actual financial results to a prior period or a budget, isolating the differences that are large enough to matter, and requiring an explanation for each one. It runs on two comparisons — period-over-period and budget-vs-actual — filtered through a materiality test so small, expected fluctuations don't bury the handful of movements that actually need management's attention. This page walks through the full process, the dual threshold that separates noise from signal, a worked example, and how accounting firms run variance analysis faster with OCTA Flow while a human reviews every explanation.

Why variance analysis matters, and where it goes wrong

Every account moves period to period — revenue grows, expenses fluctuate seasonally, one-time items come and go. Variance analysis exists to answer a specific question for every account: is this movement expected, or does someone need to explain it? Without a disciplined process, that question gets answered inconsistently. A reviewer eyeballs the P&L, notices a few numbers that "look big," and moves on — while a account that moved 40% on a small base, or one that crept up steadily below anyone's attention threshold, goes unexamined.

The two most common failure modes are opposite errors. Flagging every account that moved by some percentage buries the analysis in noise — a $2,000 line item that doubled isn't material to a $50 million company, but the review time it costs is real. Flagging only large dollar movements has the reverse problem — a $40,000 shift in a small, sensitive account (say, a legal reserve or a related-party balance) can be far more consequential than a $40,000 shift in a $10 million revenue line, and a dollars-only filter misses it. The fix is a dual test: an account is only flagged when the movement clears both a percentage threshold and a dollar threshold. That combination is what turns variance analysis into a control instead of a guessing exercise, and it's the backbone of the process below.

The variance analysis process, step by step

A rigorous variance analysis follows a consistent sequence, run every close. The steps below are the full procedure OCTA Flow executes; they also stand alone as a best-practice process any firm or finance team can follow.

1. Set the materiality thresholds. Before comparing a single number, agree on the dual test: a percentage threshold (commonly 10%) and a dollar threshold (commonly $50,000, scaled to the size of the entity). An account is only flagged when a variance clears both conditions — this is the filter that determines everything downstream.

2. Compute period-over-period variance. For every account with a prior-period comparative available, match accounts by number or name and calculate: $ Change = Current − Prior, and % Change = (Current − Prior) ÷ |Prior|. Flag any account that clears both thresholds, and note the direction — favorable (F) or unfavorable (U) to net income.

3. Compute budget-vs-actual variance. For every account with a budget or forecast available, calculate: $ Variance = Actual − Budget, and % Variance = (Actual − Budget) ÷ |Budget|. Flag accounts clearing both thresholds. Direction depends on account type: for revenue, actual above budget is favorable; for expenses, actual below budget is favorable — reversed of revenue.

4. Calculate the supporting financial metrics. Beyond line-item variances, compute the ratios that reveal structural shifts a single account can hide: revenue growth rate, gross margin %, operating expense as a % of revenue, and SG&A as a % of revenue — each compared to the prior period and to budget. A margin or expense-ratio move can be material even when no single account trips the dollar threshold.

5. Compare to industry benchmarks, where available. If benchmark data exists for the entity's sector, compare the computed ratios against it and flag any deviation greater than 20% from the benchmark. This surfaces structural issues a purely internal (period-over-period or budget) comparison can't — a margin that looks stable against last year but is well below what peers in the same industry are running.

6. Run a monthly trend check. For accounts with monthly detail available, plot the balance across the year and look for trend breaks, seasonal anomalies, and any month with a spike inconsistent with the account's annual pattern. A variance that looks unremarkable annualized can still hide a one-month anomaly worth flagging on its own.

7. Generate an explanation request for every flagged account. Each flagged item gets a formatted request — account name, current balance versus the prior-period or budget figure, the variance amount and percentage, and a prompt for the responsible party to provide a written explanation. Nothing stays flagged without a documented response.

8. Rank flagged items by priority. Sort every flagged account by absolute dollar variance, descending, and assign a priority: High for variances more than 2× the dollar threshold, Medium for variances between 1–2× the threshold, and Low for variances just clearing the threshold. Priority ranking is what tells a reviewer where to spend their limited time first.

Worked example: the dual materiality test

Say a firm sets its thresholds at 10% and $50,000 for a client's Q3 close. Three SG&A-adjacent accounts moved versus budget:

The first two accounts each fail one side of the test and are correctly left alone — office supplies moved 72% but on a base too small to matter; hosting costs moved a real dollar amount but only 3%, unremarkable for a scaling cost. SG&A is the one account that clears both thresholds, so it's the only one that generates an explanation request.

SG&A: $250,000 unfavorable variance vs. budget (25.0%). Actual $1,250,000 against a budget of $1,000,000. Because expenses running over budget are unfavorable to net income, this is flagged unfavorable. For priority: the dollar threshold is $50,000, and $250,000 is exactly 5× that threshold — well above the 2× cutoff — so this account is ranked High priority, ahead of any Medium or Low item on the same schedule, and routed for an explanation request with an option to escalate if the response doesn't hold up.

Key controls and red flags

Beyond the dual-threshold mechanics, a rigorous variance analysis watches for patterns that a simple flag/no-flag test can miss:

• Margin compression — revenue accounts declining while cost accounts rise in the same period, even if no single account trips the dollar threshold
• Gross margin declining more than 5 percentage points period-over-period or against budget
• SG&A growing faster than revenue — a sign of operating leverage moving in the wrong direction
• Expense spikes concentrated near period-end with no corresponding revenue or business driver — a common sign of cutoff issues or expense management
• Accounts with zero activity in the current period that had activity in the prior period — worth a quick check that the account wasn't simply reclassified or missed
• Budget variances greater than 25% in core revenue or COGS lines — flagged regardless of the standard dual test, because these are the accounts a board or lender will ask about first
• Ratios that deviate more than 20% from industry benchmarks, even when the trend looks stable internally

Consistently catching these — every account, every period, against the same thresholds — is what separates a genuine variance control from a once-over glance at the P&L.

What a completed variance analysis produces

A finished variance analysis isn't just a spreadsheet of percentages — it's a documented workpaper with management's explanations attached, ready for a partner or board to review. A complete variance analysis package includes:

How OCTA Flow automates variance analysis

OCTA Flow runs the comparisons and the materiality test for you, and leaves the explaining — and the sign-off — with your team. The workflow mirrors the process above:

1. Pick the Variance Analysis Skill. Flow already knows the full procedure: dual-threshold testing, period-over-period and budget-vs-actual comparisons, ratio and benchmark analysis, and priority ranking.
2. Connect your data. Point Flow at the accounting system, or upload the period's files — the current trial balance, the materiality thresholds, and whichever of the prior-period, budget, benchmark, or monthly detail files are available.
3. Run. Flow computes every variance it can — period-over-period where a prior period exists, budget-vs-actual where a budget exists — applies the dual materiality test, calculates the supporting financial ratios, and checks them against benchmarks and monthly trends where the data supports it.
4. Review findings by severity. Instead of a full schedule to scan line by line, Flow surfaces only the accounts that cleared both thresholds — ranked by priority, each with the specific figures, the variance amount and percentage, and a recommended action: request an explanation, escalate to the CFO, accept with a documented reason, or dismiss. Your team works the flagged accounts, not the entire trial balance.
5. Sign off. Once every flagged account has a management response — or a documented override — Flow assembles the workpaper with the full audit trail intact.

The result: the mechanical comparison across every account is done in a fraction of the time, and your people spend their hours writing and reviewing the explanations that actually require judgment.

Control and trust: Flow proposes, you approve

This is what matters most to a firm putting its name on the numbers: OCTA Flow never writes to your books, and it never manufactures a management explanation. Flow computes the variances, applies the materiality test, and generates the explanation requests — but every written response, override, and escalation decision comes from your team.

That control model runs through the whole analysis:

• Findings, not conclusions. Flow tells you what moved and by how much — your team decides why, and whether that reason holds up.
• Priority and escalation built in. A high-priority variance that lacks a credible explanation can be escalated to the CFO or audit committee rather than quietly accepted.
• A complete audit trail. Every computed variance, every explanation request, every management response, and every override is logged, so the analysis is fully traceable end to end.

You get the speed of automating the comparison, with the accountability of a human standing behind every explanation.

What the analysis draws on

To run a variance analysis, Flow uses the same sources a preparer already works from:

• Current-period trial balance — the account balances for the period under review (required)
• Materiality thresholds — the percentage and dollar cutoffs used for the dual test (required)
• Prior-period financial data — comparative balances for period-over-period variance (optional, but needed for that comparison)
• Budget data — approved budget or forecast amounts by account, for budget-vs-actual variance (optional, but needed for that comparison)
• Industry benchmarks — sector financial ratios, for the benchmark comparison step (optional)
• Monthly GL detail — for trend analysis on higher-risk accounts (optional)

Flow works from whichever of these your client has connected or uploaded — it matches each input by its purpose, so it runs the comparisons the available data supports and simply skips the ones it can't (for example, no budget-vs-actual test without a budget file).

Related skills and terms

Frequently Asked Questions

What is variance analysis in accounting? Variance analysis is the process of comparing actual financial results to a prior period or a budget, calculating the dollar and percentage difference for each account, and flagging the differences large enough to require an explanation. It's typically run every close as part of the review process before financial statements go out.

What is the difference between period-over-period and budget-vs-actual variance? Period-over-period variance compares the current period's actual results to the prior period's actual results — it shows how the business is trending. Budget-vs-actual variance compares current actual results to what was planned or forecast — it shows how the business is performing against its own targets. A complete analysis runs both where the data supports it.

What is a materiality threshold in variance analysis? It's the cutoff that determines whether a variance is worth investigating. A dual test — commonly a percentage threshold like 10% AND a dollar threshold like $50,000 — flags an account only when both conditions are met. This avoids flagging small-dollar swings that happen to be a large percentage, and large-dollar swings that are only a small percentage of a big account.

What is flux analysis, and is it the same as variance analysis? Flux analysis and variance analysis refer to the same core process — comparing account balances period-over-period or against budget and explaining the fluctuations ("flux"). The term "flux analysis" is more common in audit and public-company reporting contexts; "variance analysis" is the broader accounting and FP&A term.

How do you rank which variances to review first? Sort flagged accounts by absolute dollar variance, descending, then assign priority based on how far each clears the dollar threshold — commonly High for more than 2× the threshold, Medium for 1–2×, and Low for anything just at the threshold. This puts the accounts most likely to matter at the top of the reviewer's list.

Can variance analysis be automated? The comparison math, the materiality test, the ratio calculations, and the explanation-request generation can all be automated and reviewed, while writing and approving the actual explanations stays with your team. That's the model OCTA Flow uses.

Does OCTA Flow decide whether a variance is acceptable? No. Flow computes the variance, applies the materiality test, and generates the explanation request — but it never accepts, overrides, or dismisses a finding on its own. A person on your team reviews the explanation and makes that call.

See how firms run faster, fully-explained variance reviews with human sign-off → start a 30-day OCTA Flow trial.