Where budgets stop being guesswork

Most people treat budget analysis as a reporting task. Is that the problem?

Reporting is what happens after the decisions are already made. Budget analysis is supposed to inform those decisions — which means it needs to happen earlier, with more context, and with someone who understands the operations behind the numbers.

The finance team hands over a variance report in week five. By then, the damage is done. That timing issue alone explains most of the friction I see between departments.

What does a well-structured budget model actually look like?

Separate assumptions from calculations. That sounds obvious but most models I inherit have hardcoded values scattered through formulas — you change the inflation rate in one cell and miss four others.

Good structure means you can stress-test it in under ten minutes. If explaining the model takes longer than running a scenario, the model is too opaque to be useful.

Where do organisations lose control of variance?

Usually in the categorisation stage. Teams split costs inconsistently — same expense type shows up under three different line items depending on who submitted it. Then variance analysis becomes archaeology rather than analysis.

A chart of accounts that everyone actually follows, reviewed quarterly, removes roughly 60% of the reconciliation problems I used to spend weeks on.

You teach rolling forecasts at Unbuyible — why that over static annual budgets?

A static annual budget reflects what you believed in October about what would happen in August. Rolling forecasts update every four to eight weeks using actual data, so the plan at least loosely resembles current reality.

It requires more discipline and better process hygiene. But the organisations that maintain both — a fixed annual baseline and a rolling view — make noticeably better resource allocation decisions mid-year.

What should someone coming into budget analysis learn first?

Read the operations before touching the spreadsheet. Talk to the people who generate the costs — warehouse staff, procurement, whoever manages headcount. The numbers make more sense once you understand what they represent.

Technical skills like Excel, Power BI, or basic SQL matter, but they are easier to acquire than business context. Get the context first.

One thing that surprised you when you moved into teaching?

How much people underestimate the communication part. Producing a correct analysis and getting that analysis acted on are genuinely different skills. Most courses skip the second one entirely.

We spend at least a third of every module on how to present findings to people who are not analysts — what to cut, what to keep, and how to frame a recommendation without sounding defensive.