Budget versus actual is the monthly comparison of what you planned to spend and earn against what actually happened, with a written explanation for every material difference. It runs within a week of close, it covers revenue and every department, and it explains favorable variances as carefully as unfavorable ones.
Done properly it takes about an hour a month and it is the single highest-return reporting habit a startup can build. Done as a table with no commentary, it is a file nobody reads.
Why the plan is worthless without this
An annual operating plan that never gets compared to reality is a document produced in November and ignored by March. Variance analysis is what converts it into a management tool.
Three things happen when you run it consistently. You find problems in month two rather than quarter three, when they are still small enough to fix. Your forecasting improves, because explaining a miss every month makes you better at predicting the next one. And your board learns that your numbers mean something, which is the asset you draw on when you need them.
The reverse is also true, and it is common. A company that quietly edits the budget each month to match actuals has a plan that is always right and always useless, and a board that eventually notices.
Set a materiality threshold first
Not every variance needs explaining. Without a threshold, the review becomes a 40-line table of noise and the real problems hide inside it.
Set the threshold in advance and apply it consistently. A reasonable starting point for a Series A or B company is a variance greater than 10% of the line and greater than $5,000. Both tests, not either. A 40% variance on a $1,200 software line is not worth a paragraph. A 6% variance on a $300,000 payroll line is.
Write the threshold into the reporting format so nobody has to argue about it each month.
The four questions every explained variance must answer
For each variance over the threshold:
What was the gap? Plan, actual, dollar variance, percentage variance.
Why? One sentence. Not “higher than expected,” which restates the number. “Two engineering hires started in February instead of April” is a reason.
Is it timing or structural? This is the distinction that matters most, and the next section covers it.
What happens next? Either the action you are taking, or an explicit statement that no action is needed and why.
If a variance cannot get a real answer to the second question, that is itself the finding. It usually means the coding is wrong or someone spent money nobody tracked.
Timing versus structural, the distinction that matters
A timing variance means the money moves between periods but the year-end number holds. An annual insurance premium paid in March instead of April. A conference invoice that landed a month early. A hire starting late.
A structural variance means your assumption was wrong and the full-year number changes. Cloud costs running 40% above plan every month because usage scaled differently than modeled. A sales rep ramping to quota in nine months instead of five.
The reason this matters: timing variances need a note, structural variances need a decision. A company that treats every miss as timing ends the year 20% over budget, having explained away the same overrun twelve times in a row.
The test is simple. Ask whether this variance will reverse itself within the next two months. If the honest answer is no, it is structural, and something in the plan needs to change.
Favorable variances deserve the same treatment
Underspending is not automatically good news, and reporting only the misses teaches your board that the beats were luck.
Marketing coming in 30% under plan usually means the campaign did not launch, not that someone found a discount. Engineering payroll under plan means hiring is behind, which is a pipeline problem showing up in the finance report before it shows up anywhere else. Revenue over plan is worth understanding precisely, because you want to know whether to do more of whatever caused it.
The rule: same threshold, same four questions, both directions.
What common variance patterns actually mean
A few recur often enough to name.
Payroll consistently under plan. Hiring is behind. Check the recruiting pipeline before celebrating the saving, because the missed hires usually show up as a revenue or roadmap problem two quarters later.
Payroll over plan with headcount on plan. The loading was wrong. Base salary is roughly 75% to 80% of true cost once employer taxes, benefits, and bonus are included, and a plan built on base alone runs over every month.
Cloud and infrastructure creeping up monthly. Almost always structural. Usage scales with customers and with engineering behavior, not with your budget.
G&A over plan. Usually legal, insurance, or a renewal nobody had in the model. D&O premiums in particular jump when an institutional board is added. Our G&A budget guide covers realistic allocations by stage.
Revenue on plan but collections behind. Not a revenue variance at all. It is a cash problem and it belongs in your 13-week cash flow forecast, where it will show up sooner.
Building the report
Structure it in this order, because it is the order a reader wants.
- Summary line. Revenue, total spend, and EBITDA or net loss: plan, actual, variance, for the month and year to date.
- Revenue detail, split into new, expansion, and churn so the miss can be located.
- Spend by department, each with its own plan and actual.
- Variance commentary, one short paragraph per item over threshold.
- Cash and runway, with actual runway against what the plan projected.
- Forecast impact. Given what you now know, where does the full year land?
Both month and year-to-date columns matter. A single bad month is noise. The same variance three months running is a trend, and year to date is where it becomes visible.
This section is a standing part of the board package. Our guide to what goes in a startup board reporting package covers the rest of the format.
When you are consistently off
If you are missing the plan in the same direction three months running, the problem is the plan, not the month.
Two honest responses, and they are different.
Re-forecast. Update your expectation for the rest of the year based on what you now know. Do this quarterly, and present it alongside the original plan rather than instead of it. The board should be able to see both what you committed to and where you now expect to land.
Do not rewrite the budget. The approved plan stays frozen as the baseline. A company that revises the budget to match actuals loses the ability to measure anything, and the variance report becomes a formality.
If the gap is large enough that the plan is no longer a useful baseline, that is a board conversation, not a spreadsheet edit.
Who runs it
Finance produces the report and the commentary. Department leads own their own variances and should be able to explain them without finance translating.
That last part is what makes the process work. If your VP of Sales cannot explain why their line came in 18% over, they are not managing to the number. The variance review is where budget ownership actually gets tested, which is exactly why it is worth doing in a meeting rather than over email.
At most Series A and B companies the person producing this is a controller or fractional CFO rather than the founder. Our comparison of fractional CFO vs. controller vs. VP Finance covers where the line falls.
Download the model
The 2027 budget model we use with clients is free to download, no email required. Eight linked tabs including a full P&L, balance sheet, and cash flow with monthly burn and runway.
To use it for variance reporting, save a copy of the approved plan and keep it frozen. Track actuals in a parallel copy, and compare by department month by month. Freezing the original is the whole point.
Common questions
How often should we run budget vs. actual?
Monthly, within a week of close. Quarterly is too slow to catch a problem while it is still small, and weekly is more precision than monthly accounting can support.
What variance threshold should we use?
Greater than 10% of the line and greater than $5,000 is a reasonable starting point for a Series A or B company. Both tests together, applied consistently, and written into the format.
Should we explain favorable variances too?
Yes. Underspending usually means something did not happen, and reporting only the misses trains your board to discount the beats.
What if the budget was just wrong?
Then say so, explain what you now believe, and re-forecast at the next quarterly checkpoint. Leave the original plan frozen as the baseline. Being wrong and saying so plainly costs far less credibility than adjusting the plan until you are always right.
Is budget vs. actual the same as variance analysis?
Practically, yes. Budget vs. actual is the comparison. Variance analysis is the explanation attached to it. The comparison without the explanation is a table, and the table is not the useful part.
How long should the variance commentary be?
One short paragraph per item over threshold. If it runs longer, the issue is big enough to be its own agenda item rather than a footnote in a report.
If your monthly variance review is a scramble, or your board is asking questions your reporting cannot answer, that is fixable. Start a conversation.