November offer: EXTRA 30% off all courses & bundles. Use NOVEMBER30 at checkout. Ends 30 Nov.

Budget vs Actual Variance Analysis: Explain the Difference

May contractor costs are $15,000 over budget. Compare the month with YTD and trace the postings before you change the forecast.
Nov 21 / Peter Ilas, FCCA
That is where budget vs actual variance analysis earns its place in the monthly review. My first check is to put the month beside year to date (YTD). If they tell different stories, I want to see the postings before I tell management what changed.

The aim is not to write a longer comment. It is to give the reader a sound explanation and make the next action clear.

The short answer: investigate before explaining

The budget variance formula is Actual minus Budget. Compare monthly and year-to-date results on the same basis, then trace the difference to the postings before explaining it.

A monthly overspend can be a timing difference

Consider a fictitious contractor-cost example from my course. All figures below are in USD thousands.

Fictitious contractor costs (USD thousands)

PeriodActualBudgetVariance
May5540+15
January–May2002000
May is 15 over budget: (55 − 40) ÷ 40 = 37.5%. On its own, that can sound like a continuing cost increase. But the YTD figures tell a different story: the total is exactly on budget.

For a positive budget, divide the variance by the budget and multiply by 100. If the budget is zero, report the variance amount without a percentage. If the budget is missing, flag the comparison as unavailable.

The earlier review note says a missing April invoice of 15 was expected in May. In this illustrative case, that is the context behind the movement. In a live review, I would match the invoice to the May posting and check whether April already included an accrual. If it did, I would look for the reversal before concluding that all 15 belongs in the explanation. An old note tells me where to look; it does not prove what was posted.

The invoice crossed months

AprilInvoice missing

The earlier review note says to expect the $15,000 invoice in May.

MayInvoice arrives

Match it to the posting and check for an April accrual or reversal.

The earlier note points to timing. Match the invoice to the posting before calling it a timing difference.
A better comment could be: “May contractor cost is 15 above budget following receipt of an April invoice. January–May cost is in line with budget.” Use that wording only once the posting and timing have been verified. If the evidence is incomplete, say what needs to be checked instead of writing the conclusion as fact.

The point is not that YTD agreement makes everything correct. It is that a monthly percentage alone cannot establish the cause.

Keep the comparison basis consistent

Before investigating a difference, confirm the entity, account scope, business unit, cost center, currency and period. Use the same dimensions for Actual, Budget and prior year.

A supplier search may help find the relevant invoices. It does not make supplier-only Actual comparable with the full department budget. Likewise, missing Budget data is not automatically zero; labelling it as zero can manufacture a misleading variance.

Compare like with like

Different scope

ActualOne supplier
BudgetWhole department

Same scope

ActualWhole department
BudgetWhole department
Use the same scope on both sides before you interpret the difference.
Separate the arithmetic from the financial interpretation. A positive cost variance normally means higher cost than budget. A positive revenue variance normally means higher revenue. Whether a variance is favorable or unfavorable depends on the line, not the sign alone. Under Actual minus Budget, the same positive difference usually helps profit on a revenue line and hurts it on a cost line.

How to do budget vs actual variance analysis

For a meaningful variance, I would follow this route:
  1. Identify the Monthly and YTD movement within the same scope.
  2. Open the relevant postings and inspect dates, descriptions, amounts and supporting records.
  3. Read the trend and previous-month comments for an expected catch-up, reversal or change in run rate.
  4. Check an apparent explanation against the current evidence.
  5. Record the conclusion, remaining question and responsible follow-up.
An annual software charge, a reversal and a new recurring cost can all produce a large monthly movement. They should not 
receive the same explanation. A highlighted cell shows where to investigate; it does not settle the interpretation. The same principle applies when you build a direct cash flow from bank data: the total can agree and still tell you little about what moved it.

Write for the person making the decision

The cost-center owner may need the detailed posting explanation. A business-unit manager needs the main drivers and actions. A company-level reader usually needs the material conclusion, not a copy of every lower-level note.

Keep Monthly and YTD comments distinct when the stories differ. In the contractor example, the month has a timing catch-up while the YTD total remains on budget. Collapsing both into “contractor costs overspent” loses an important distinction.

Good variance commentary should answer: what moved, why, what evidence supports that explanation and what happens next? If the cause is not yet confirmed, make that uncertainty visible.

Use AI for drafting—not automatic acceptance

In the course workflow, AI has two separate roles. It helps build and improve the local review tool. A separate AI conversation can also check existing explanations or propose new commentary using approved inputs and earlier comments.

The dashboard runs locally. The AI conversation is a separate drafting and review step, not an assistant embedded in the dashboard.

Useful instructions ask AI to keep the comparison scope clear, distinguish Monthly from YTD, avoid unsupported causes, follow the intended audience and identify missing evidence. Prior writing can guide style, but every new explanation still needs review.

Before accepting a draft, check the numbers, direction, timing and stated cause. A plausible reference to a delayed hire, for example, still needs evidence of the actual start date and pay. Do not interpret an AI follow-up request as work already performed.

A draft is not evidence

Draft

AI proposes wording from approved inputs.

Verify

Finance checks the numbers and the cause against evidence.

Decide

Accept or revise. Keep open questions visible.

AI can draft the explanation. Finance checks the evidence and decides what to use.
Use fictitious or approved data and follow your organisation’s confidentiality and AI policies. Local use of the finished tool and sending information to a separate AI service are different data-handling decisions.

Continue the review when the numbers change

A recurring review needs more than a clean first dashboard. New postings may change a prior explanation, and the next month should not begin with lost comments or unresolved questions.

Keep the reviewed scope, explanations and open actions together. When data changes, inspect the effect on the previous conclusion. Preserve the prior context and explicitly acknowledge what was reviewed rather than silently carrying forward an outdated statement.

Before carrying a comment forward, ask what would make it wrong. If a supposed one-off appears again the next month, reopen the explanation. If the invoice is still unmatched, keep the question open and name who will check it. Neither decision should disappear when you refresh the file.

A checklist for the next review

  • Are Actual, Budget and prior year comparable for this scope?
  • Have Monthly and YTD both been considered?
  • Can the explanation be traced to postings and supporting evidence?
  • Is a timing difference being mistaken for a recurring change?
  • Does the wording distinguish confirmed facts from follow-up questions?
  • Will comments and unresolved work remain available next month?

Key takeaways

  • Read monthly and YTD results together before treating a variance as a run-rate change.
  • Match the explanation to postings and supporting evidence, not just to last month’s note.
  • Use AI for a draft; keep the decision and any open questions with the finance reviewer.

FAQ

Does a large percentage variance mean a serious problem?

Not necessarily. Consider the amount, the comparison base, timing and business context. A small budget or missing comparison data can make a percentage especially misleading.

Does being on budget YTD remove the need to investigate?

No. Offset movements or timing can conceal important differences. Use YTD as context, not as proof that the underlying postings and explanations are correct.

Can AI write the management commentary?

It can propose and check wording using approved information. The finance professional remains responsible for checking the figures, supporting evidence and conclusion before acceptance.

Build the workflow, not just the variance table

In my course, AI Finance: Build an AI-Powered Budget vs Actual Review Workflow, you build a local review tool without writing code, investigate actuals against budget and prior year, and practise drafting and checking commentary with AI. Then you carry the review forward as new figures arrive.

The course is designed for 6.0 CPE credits. Confirm acceptance with your provider or professional body.

Peter Ilas, FCCA

Founder, RSAgile | E-LITE Consulting
30 years in finance, accounting, audit and internal controls.
Builds practical AI tools for finance teams.
Created with