A variance is actual minus budget, expressed in dollars and as a percentage of budget. Useful variance analysis adds three things: the direction, favourable or unfavourable; a materiality threshold so only real movements get investigated; and a named action.
How to calculate a variance
A variance is actual minus budget. Divide that by the budget figure and multiply by one hundred for the percentage variance, which is the number you should sort your report by. Report both: a large percentage on a small line is noise, and a small percentage on your biggest line is usually the thing worth talking about.
Only one side of that subtraction is fixed. Actuals come out of the financial records every company has to keep under section 286 of the Corporations Act 2001. The budget is nothing but judgement, written months earlier by people who did not know what would happen.
The sign of the number is not the same as its direction. A variance is favourable when it improves profit and unfavourable when it reduces it, so revenue above budget and expenses below budget are both favourable even though one is a positive difference and the other negative. Label the column favourable or unfavourable rather than leaving readers to work out the sign convention, because half of them will get it wrong.
Revenue variances and expense variances behave differently
They are not the same analysis and should not be presented as one table. A revenue variance is usually a mix of volume and price, and separating those two is the first thing to do, because selling fewer units at a higher price is a very different business event from selling more units at a discount, even where the revenue line matches.
An expense variance is more often a timing question than a spending question. A contractor invoice that lands in April instead of March produces an unfavourable variance in one month and a favourable one in the next, and neither is real. Checking whether a variance reverses next month is the cheapest filter available, and it removes most of the items that would otherwise consume the review meeting.
| Type of variance | What to look at before you write commentary | |
|---|---|---|
| Revenue | Actual sales against budgeted sales | Split the movement into volume and price before writing anything. Fewer units at a higher price and more units at a discount can produce the same revenue line and mean opposite things. |
| Direct cost | Cost of delivery against budget | Read it against actual volume, not budgeted volume. A cost overrun on a month where you sold more than planned is often not an overrun at all. |
| Headcount | Payroll against plan | Usually a start-date variance rather than a salary variance. Check against the hiring plan first, because a favourable payroll line often means a role you needed has not been filled. |
| Timing | Anything that reverses next month | Test every unfavourable expense variance for this before investigating it. If it reverses next month, the commentary is one line and the meeting moves on. |
Materiality: which variances deserve investigation
Most variance reports fail because they investigate everything, which means they investigate nothing well. Set a threshold before the period closes, expressed as both a dollar figure and a percentage, and investigate only what breaches either. State the threshold on the report so a reader knows what silence on a line means.
The cost of getting this wrong is not just time. When volatility makes forecasts change week to week, finance teams end up spending more time explaining variances than delivering strategic value.[1]That is the failure state: a function fully occupied narrating the past. A threshold is the mechanism that buys the time back.
A large variance is sometimes a budget problem rather than a performance problem. Forecast accuracy is itself a measurable capability, and improving it is a genuine achievement finance leaders track, for instance moving forecast accuracy from 50% to 95%.[2] If the same line breaches the threshold every month in the same direction, the budget is wrong and re-forecasting beats explaining.
Variance commentary sits inside the wider reporting pack, covered in what founders get wrong about the profit and loss statement.
Turning the analysis into action
A variance report earns its place when every material line carries three things: what moved, why it moved, and what happens now. The third is the one that gets dropped, and without it the board invents its own reason for the movement, which is worse than not showing it. Once that commentary leaves the building in an investor update, ASIC's guidance on disclosing financial information that sits outside the accounting standards is a sensible bar to write to, because budget-versus-actual figures are exactly that kind of number.
There are only four honest endings to a material variance. The forecast is updated because the change is permanent. Spend is corrected because the movement was a decision that should not have been made. Nothing is done because the variance is timing and will reverse. Or the budget is rebuilt because the assumption behind it no longer holds. Every line should resolve to one of those.
Who owns variance analysis
Producing the numbers and explaining them are different jobs. A financial analyst provides the detailed analysis of budgets and forecasting and prepares the FP&A reports; senior finance then provides the commentary and recommendations on top.[3] In a small team one person does both, which works until the volume of explanation squeezes out the analysis.
This is also the part of finance most exposed to automation. It comes up constantly in the conversations I have with finance leaders, including the ones I record for The CFO Track. Based on those conversations, finance teams will get leaner while expectations grow: where AI tools can generate reports, flag variances or draft commentary, leaders will opt for smaller teams with stronger commercial influence.[4] That should change how you hire for this work. The person who can assemble the variance table is becoming cheaper to replace. The person who can sit with an operations lead and work out why the number moved is not.
The role that usually carries this work is set out in what an FP&A analyst actually does.
Common questions
How do you calculate variance in budgeting?
Take actual minus budget for the period, then divide by the budget figure and multiply by one hundred for the percentage. Show both numbers, then sort by the dollar column rather than the percentage one. Sorting by percentage puts a four hundred dollar stationery overspend above a forty thousand dollar miss on your largest cost line, which is how variance reports end up being read from the bottom.
What is the difference between a favourable and an unfavourable variance?
A variance is favourable when it improves profit and unfavourable when it reduces it, which is not the same as whether the arithmetic difference is positive or negative. Revenue above budget and expenses below budget are both favourable, so the sign changes meaning halfway down the page. The exception worth knowing is that favourable is not automatically good news: a favourable payroll line usually means a role you budgeted for has not been filled, which is a delivery problem wearing the costume of a saving.
Which variances should you actually investigate?
Only the ones that breach a threshold you set before the period closed, expressed as both a dollar amount and a percentage. A threshold set afterwards is a justification, and everyone in the room knows the difference. State the number on the report itself, so that silence on a line means something specific. Then take one pass for timing before investigating anything, since an invoice that lands a month late creates a variance in two consecutive months and neither one of them is real.
Who should own variance analysis in a growing company?
Producing the numbers and explaining them are different jobs, and only one of the two is worth hiring for now. Assembling budget against actuals into a report is the part software is getting good at fastest. Sitting down with an operations lead, working out why the number moved and deciding what changes as a result is not, and it is also the part nobody can do without knowing the business. Hire for the second and let the first shrink.
References
- A pattern I keep seeing: when global headwinds create volatility and forecasts change week to week, finance teams spend more time explaining variances than delivering strategic value.
- From a recent search: it is worth putting metrics to achievements, such as showing forecast accuracy improved from 50% to 95%, to illustrate the level of impact and value a candidate brought to a role.
- How I explain it: a financial data specialist who provides detailed analysis of financial data including budgets and forecasting, and prepares FP&A reports for senior finance teams to provide commentary and recommendations.
- Where I think this is heading: finance teams will get leaner but expectations will grow, and where AI tools can generate reports, flag variances or draft commentary, leaders will opt for smaller teams with stronger commercial influence.