Variance analysis: turning budget-vs-actual into decisions

Every month you can produce a report that lays your actual results next to your budget. Most companies do produce it, glance at it, and file it. That is a waste, because the point of the exercise is not the report; it is the handful of decisions it should trigger.

Variance is just the difference between what you planned and what happened. Variance analysis is the short, honest conversation about why the big differences occurred and what, if anything, you are going to do about them. The skill is not in the arithmetic. It is in ignoring almost everything and concentrating on the few numbers that actually move the business.

Reading a simple budget-versus-actual table

Here is one month for a small company. Read across: budget, actual, variance, and whether the variance helps or hurts.

Revenue: budget $200,000, actual $182,000, variance $18,000 short, unfavorable. Cost of delivery: budget $80,000, actual $79,000, variance $1,000 under, favorable. Payroll: budget $70,000, actual $77,000, variance $7,000 over, unfavorable. Marketing: budget $15,000, actual $9,000, variance $6,000 under, favorable. Software: budget $4,000, actual $4,300, variance $300 over, unfavorable.

Now stop and notice what you do not need to discuss. The $1,000 on delivery and the $300 on software are noise. Nobody should spend a minute on them. Three lines matter: revenue down $18,000, payroll up $7,000, and marketing down $6,000. Everything else can wait.

Focus on the few that matter

A useful rule of thumb: only investigate variances that are both large in dollars and large in percentage. A line that is off by 40 percent but only $300 is irrelevant. A line off by 9 percent but $18,000 deserves a real explanation. Set a threshold so you are not chasing rounding, for example anything over $5,000 and over 5 percent.

Then explain each survivor in one plain sentence. Not a paragraph, not a spreadsheet of sub-causes. One sentence a non-finance person would understand:

• Revenue down $18,000: two deals we expected to close in January slipped to February.
• Payroll up $7,000: we backfilled the support role a month earlier than budgeted.
• Marketing under $6,000: we paused the paid campaign while we rebuilt the landing page.

Decide what changes

An explanation is not the end. Each variance points to one of three responses: do nothing, change the plan, or change behavior. The discipline is saying which, out loud, for each one.

Take the three above. The revenue slip is timing, not loss, so the answer is do nothing this month but confirm those deals land in February. The early payroll hire was a deliberate choice that will repeat, so the answer is change the plan: payroll is now permanently $7,000 a month higher than budget, and that flows into your forecast for the rest of the year. The paused marketing freed $6,000, so the decision is whether to redeploy it or let it drop to the bottom line. Notice that timing differences and permanent differences need completely different treatment, and confusing the two is the most common mistake in this work.

Make it a habit, not an event

This should take thirty minutes a month, not a day. The same people, the same short table, the same three questions: which variances are big, why did each happen, and what changes. Write the one-sentence explanations down so that next month you can see whether last month's promised actions actually happened.

Done consistently, variance analysis quietly tightens the whole operation. You stop being surprised by your own numbers, your forecast gets more accurate because you keep folding real causes back into it, and when an investor or lender asks why a line moved, you have a calm one-sentence answer ready instead of a guess.