Variance analysis interview questions: one worked bridge

Explore the Finance case interviews hub and the finance careers guide.

Variance analysis interview questions sound like math. They're mostly about order.

Anyone can subtract actual from budget. The interviewer is checking whether you can split the gap into causes that add back to the total, say which ones come back next quarter, and say what to do about it.

Below is one revenue miss, worked through with our own numbers, then the questions around it. The rest of the interview loop is in the finance interviews guide.

What are variance analysis interview questions really testing?

Variance analysis interview questions test whether you can explain a number. The subtraction takes a second. The interviewer wants the gap split into causes, a check that the causes add up, a view on which ones repeat, and a recommendation, which is the job an FP&A seat does every month.

Whether it's "Revenue came in 7 percent under budget, walk me through it" or a take-home table, the answer has one shape.

The behavioural version is on the FP&A interview page.

What is variance analysis?

The Corporate Finance Institute puts it simply: "Variance analysis is the analysis of the difference between planned and actual numbers." A variance is favourable when it helps profit and unfavourable when it hurts it. Revenue over plan is favourable, and a cost over plan is unfavourable.

That sign flip is the first trap, so you say "favourable" or "unfavourable", never "up" or "down".

Textbook lists split factory costs into price and quantity. In FP&A interviews you'll mostly get revenue or operating expenses, split into volume, price, mix, currency and one-time.

Budget vs forecast vs actual: which variance are they asking about?

A budget is the plan the company commits to for the year. A forecast is the latest estimate of where the year will land. So actual versus budget asks whether the plan held, and actual versus forecast asks whether the latest view was right. Say which one you mean before you give a number.

Steve Morlidge at FP&A Trends puts it well: "The best way to think about forecasts is that they are future actuals." Budgets, he writes, "reflect your aspirations."

The one people forget is this forecast against the last one. In the job it's usually what a leader asks about first, because it shows what moved since you last spoke.

A routine I took over was built around it. Each cycle we'd "reconcile the old version versus the new version to figure out the differences."

How a budget becomes a negotiated number is in the FP&A answer on budget versus forecast.

A worked variance analysis example: one revenue miss

This is a worked revenue miss in the shape interviewers use. One product line, the second quarter of four. Budget was 20,000 units at $600, so $12.0 million. Actual revenue came in at $11.14 million, $0.86 million under budget, about 7.2 percent unfavourable.

Underneath, 19,000 units sold at an average $610, measured at budget exchange rates. Currency cost $0.25 million at actual rates, and one customer got a $0.2 million credit after an outage.

Add the four steps in thousands: minus 600, plus 190, minus 250, minus 200. That's minus 860, the whole miss, nothing left over.

That last check is the one I held the team to.

Before anything went up the chain in a forecast week, every bridge had to tie to zero: where we started, what changed, where we landed.

If one didn't, I'd ask which ones "haven't tied" and we'd stay on it until they did.

A leftover number means one of your causes is wrong, or there's a cause you haven't found.

The other habit is how I build it. "On the bridge I need things linked not hardcoded."

Then it survives the next refresh, and you can trace every number back.

Is the variance structural or one-time?

A structural variance repeats every period until something changes. A one-time variance doesn't. In our case, the lost volume and the price increase are structural, the $0.2 million credit is one-time, and currency depends on where rates go. Only the structural part belongs in the forecast for the rest of the year.

Volume and price together are $0.41 million unfavourable a quarter, so with two quarters left that's another $0.82 million if nothing changes.

If rates stay where they are, currency adds $0.5 million more, and the rest of the year is $1.32 million worse than budget before anyone acts.

The one-time credit doesn't go in, unless it isn't really one-time.

"One-time" is the label people reach for when a number is inconvenient, so test it. The SEC's test for public companies helps: a charge shouldn't be called "non-recurring, infrequent or unusual" when it's "reasonably likely to recur within two years or there was a similar charge or gain within the prior two years" (SEC interpretation 102.03). If this customer got a credit last year too, it's a pattern.

In a subscription business I supported, "because it was subscriber based, by q one, q two you pretty much knew what the story was in terms of forecasting."

So a late-year miss on a line that predictable is real news, and one-time items live in the messier lines.

Money you didn't spend because something slipped usually gets spent later.

What's the so what?

The so what is what the variance means for the year and what you'd do about it. For our case, price is holding, volume isn't, and the rest of the year is $1.32 million short of budget if nothing changes.

That's the part that separates a manager answer from an analyst one.

Getting from the what and the why to the so what is a habit I had to coach, and that story is on the FP&A interview page.

Out loud, the full answer runs like this. Revenue missed budget by $0.86 million, about 7 percent. Most of it is volume, partly offset by price, with currency and one credit making up the rest. The credit won't repeat, so if volume holds here the year lands about $1.32 million short. I'd find where those units went and update the forecast now.

More variance analysis interview questions and answers

Most variance analysis interview questions are the same handful asked in different words. These come up most, each with a short answer, and all of them lean on the bridge above, so learn that case first and these get easier.

"Explain variance analysis to someone outside finance." It's the gap between what we planned and what happened, split into reasons, so we know which to worry about.

"What's the difference between price, volume and mix?" Volume is how many you sold, price is what you got for each, and mix is which products sold more when they carry different prices or margins.

"How do you handle currency?" Constant rates first, currency as its own step, and the same method in every report. I once found a revenue view still on the old method.

"What was your forecast accuracy?" Be honest, even without a target. How I'd answer that is on what an FP&A manager does.

"When is a variance big enough to explain?" When it would change a decision. The controller interview page covers that materiality line.

"What would you do if the bridge didn't tie?" Find the leftover before you present anything. It's a cause you haven't named or a number in the wrong place.

How do you practise variance analysis in one evening?

To practise variance analysis in one evening, bridge one real report you know, or the case above, on paper, and make it tie to zero. Then say it out loud in under two minutes: total, causes, structural or not, full year, what you'd do.

Then change one input, like making the credit happen twice, and watch the full-year number move.

For the other common case, the break-even page has two worked cases.

Bridge it, tie it, then say what it means.