Financial reporting in Power BI: comparing actuals, budget and cost centres
Analyse actual-to-budget variances by cost centre in Power BI with a fictional example, formulas and practical checks.
A monthly report should do more than show where costs differ from budget. It should help teams ask the right question: is the difference a one-off effect, a timing shift, or a development that could affect the year-end result? To answer, actuals and budget need the same business context and an explanation with a traceable basis.
Define the rules before comparing actuals and budget
Before analysing the figures, agree the period, company, currency, cost centre, cost type and budget version. For example, compare the booked monthly value with the budget for the same month. Using an annual budget as the basis for a single month would answer a different question. For cumulative figures, both sides must cover the same period.
In the cost example below, absolute variance = actual − budget and percentage variance = (actual − budget) / budget × 100. All costs are shown as positive amounts. A positive variance therefore means spending above plan and is initially an unfavourable cost variance. Revenue and other sign conventions need their own agreed interpretation.
Fictional example: external services by cost centre
The figures, cost centres and explanations below are entirely fictional and do not represent customer results. Each comparison uses the same month and the same approved budget version; amounts are in euros.
- IT cost centre: budget €100,000, actual €120,000. The variance is +€20,000, or +20%.
- Sales cost centre: budget €40,000, actual €36,000. The variance is −€4,000, or −10%.
- Total: budget €140,000, actual €156,000. The variance is +€16,000, or approximately +11.4%.
The total percentage is calculated from the total amounts: €16,000 / €140,000 × 100. It is not the simple average of the two cost-centre percentages. The total shows the budget overrun; the breakdown reveals that lower spending in Sales partly offsets the IT variance.
Suppose the controlling team checks invoices and confirms with the IT owner that €15,000 relates to a project brought forward and €5,000 to additional services. That review provides the basis for an explanation. Whether the earlier project results in lower spending later, and how the additional services affect year-end, must be assessed separately. The monthly variance alone does not answer those questions.
From variance to decision
- Check plausibility: Do the posting period, cost-centre allocation, currency and budget version match? Are reclassifications or missing postings relevant?
- Clarify the drivers: Discuss unusual items with their owners and distinguish evidenced causes from open assumptions.
- Document the explanation: Record the cause, period and responsible person in the relevant report context. Note when clarification is still pending.
- Assess the consequence: Determine whether an action or forecast update is needed. A variance alone is not a forecast.
When budget is zero, the percentage formula is undefined. Agree an explicit display for that case, such as “not calculable”, while continuing to show the absolute amount. Negative budgets and cost or revenue sign conventions also require an agreed business rule.
Power BI data model prerequisites
For a reliable comparison, actuals and budget must be analysable using the same business keys and periods. This requires consistent cost-centre and financial-line mappings, appropriate filters and defined measures. Also check how organisational changes, missing mappings and different levels of granularity affect the comparison. Document the definitions so report readers can understand them.
Aivana Financial Reporting for Power BI compares supplied actual and budget values and their absolute and percentage variances in a financial matrix. Comments can be recorded in the context of report filters and business keys. Persistent comment storage requires a separately configured Fabric SQL connection and appropriate permissions. The visual does not replace the data model or business review of the cause. Forecast values must be supplied; planning and writeback are extension points.