An annual budget fails for process reasons far more often than for modelling reasons. The formulas work, the totals foot, and by March the file is sitting untouched on a shared drive. The revenue target was set without a bridge from last year, the department submissions arrived as numbers without a stated basis, the consolidated total missed the growth target by a wide margin and the gap was closed with an adjustment line, and the whole file was re-cut every month so nobody could be measured against it.

The model structure is a separate problem, and it is covered elsewhere. What this guide deals with is the process around the model: the calendar, the targets, the templates, the consolidation, the review round, the lock, and the ownership that keeps the numbers meaningful for the twelve months that follow.

The test of a good budgeting process is simple. In month nine, can the finance manager open the budget file, compare it to the actuals, and explain the variance to a department head who accepts the answer? If the budget has drifted, been re-cut, or quietly abandoned, the process failed long before month nine.

Working the Budget Calendar Backwards

Start with the board approval date and count backwards. A calendar that runs ten to twelve weeks gives departments enough time to think about assumptions and gives finance enough time for two consolidation passes. Compressing it into four weeks produces a set of numbers that has been filled in rather than built.

Two dates in that table do the most work. The submissions due date is a hard deadline, because a late department holds up the whole consolidation and the challenge meeting loses its evidence. The review round is where the process either produces a real plan or produces a formality, and it depends on the first consolidation being finished on time.

Top-Down Targets and the Revenue Bridge

The process starts with a growth target set by the owners or the board, then a bridge that shows how last year's revenue becomes next year's number. A bridge is far more useful than a percentage, because it names the source of the growth and gives each line an owner.

The bridge totals 9,050, which is 650 above last year and a 7.7 percent increase. Every line has an owner who can defend it, and the total ties to the target the board approves. When a department later submits a revenue plan on a different basis, the conversation stays on one bridge line and the person who owns it.

The cost envelope works the same way. Finance converts the target into a cost envelope per department, usually by applying a wage and price increase to the base and adding a small allowance for growth-related spend. The envelope is a guide for the submission, and any request that runs above it needs a reason in the challenge meeting. The margin implied by the target revenue and the total envelope is the number the board is really approving, so it belongs on the cover of the consolidated file.

The Department Submission Template

One template, issued once, used by every department. The alternative is nine files with nine layouts and a consolidation that takes a week.

A few template rules remove most of the rework. Lock every cell except the input columns, so the layout cannot be rearranged. Carry the submitter name and a version tag in named cells on the cover tab. Keep the assumptions on the tab they affect, so a reviewer reading the headcount tab can see why each role starts when it does. Require the submission to reconcile to the department's own management accounts for the current year, because a plan built on numbers that do not match the ledger will not survive the first variance report.

Consolidating Submissions Into One Workbook

Consolidation reads every submission into one staging table with a source column, then maps the accounts once. Where the files arrive in a consistent layout, Power Query can append them all in one step and refresh as revised versions land. Where they do not, a staging sheet with a department column and a version column does the same job with more manual work.

The staging table needs department, account, version and amount at a minimum. A SUMIFS pulls a department and account total into the consolidated model:

=SUMIFS(Stage[Amount],
        Stage[Department], $B$4,
        Stage[Account],    $A12,
        Stage[Version],    "v0.9")

The top-down target sits on the same sheet, pulled by INDEX-MATCH so a renamed or reordered department list does not silently break the comparison:

=INDEX(Targets[Envelope], MATCH($B$4, Targets[Department], 0))

The variance column is then the SUMIFS total less the INDEX-MATCH figure, with a check row underneath that flags any department where the imported total does not equal the total in the submitted file. Do not accept a department into the consolidated version until that check clears, because a submission that does not tie to its own file will not tie to the budget either.

Three things keep the consolidated file usable. One chart of accounts for the whole group, with the mapping tab holding the only copy of it. One workbook holding every version, so v0.1 and v0.9 can be compared without hunting through email. One version tag in a single named cell that every staging row references, so re-importing a revised submission does not overwrite the earlier pass.

The Review Round and the Challenge Meeting

The review round exists to convert submitted numbers into explained numbers. A department that lands more than a set percentage away from its target, five percent is a workable threshold for most SMEs, provides a written explanation covering the drivers, the timing and what would have to change for the target to be met. A variance with no explanation goes to the challenge meeting as an open item.

A good explanation reads like a plan and names the mechanism. "Labour is 180,000 above envelope because two service roles start in July where the earlier draft had them starting in September, and the second is a backfill for a resignation in May." A weak explanation restates the number or attributes it to growth, and both are visible in a meeting.

The challenge meeting works best run department by department, with the department head presenting the three largest variances to target and answering questions on them. The output is a list of agreed changes, each with an owner and a date, and a small number of open items that go to a second round. Keep the meeting to the numbers that move the group result, and let the small lines stand as submitted.

Locking the Budget, Ownership and Sign-Off

Once the board approves the budget, the file is locked. Version 1.0 becomes read-only, the approved version is archived beside the board papers, and the phasing is loaded into the monthly board pack so actuals can be compared to the approved plan from the first month of the year.

The version discipline that follows is what makes the budget useful. The budget stays fixed for the financial year. The rolling forecast absorbs new information as it arrives, usually as a reforecast at the half year and a shorter quarterly update. Actual against budget answers the question of how the plan is tracking, and actual against forecast answers the question of what the latest expectation is. Reporting both keeps the baseline intact and gives the board the current view.

Ownership runs line by line. The sales lead owns revenue, the operations lead owns cost of delivery, each department head owns their operating costs and their headcount plan, and the finance manager owns the consolidation, the chart of accounts and the phasing. The sign-off sits on the cover tab as a small block of names and dates covering the department head, the finance manager, the managing director and the board resolution date. A line with no owner is a line nobody will explain in month six, and the sign-off block makes the gap visible before the budget is approved.

Failure Modes That Empty the Process

Sandbagging. A department submits a number it knows it can beat, or a padded cost base it expects to be cut. The counter is a stated basis for every assumption, a comparison against the top-down envelope, and a challenge meeting that asks about the largest variances.

Six hundred line items nobody reviews. A budget detailed far beyond the level at which decisions are made gets approved in full and ignored in practice. Aim for the level at which a manager can actually control the spend, and put the rest in a grouping line.

No owner. Where the finance manager writes every department's numbers, the department heads have no stake in the result and the first variance report is a surprise to everyone. The submission has to come from the person who will be measured against it.

A budget treated as a forecast. Re-cutting the approved budget every month produces a document that always agrees with reality and measures nothing. Keep the budget fixed, forecast separately, and report the two side by side.

Bottom-up submissions with no link to the growth target. Nine departments each submit a careful plan, the consolidated total misses the board's target, and the gap gets closed with an adjustment line that nobody owns. Comparing the first consolidation to the envelope by department, before any plug is applied, is what keeps the two ends of the process connected.

A Sequence That Holds

Fix the process in this order. Publish the calendar first, working backwards from the approval date, and hold the submissions deadline. Write the target as a revenue bridge with an owner on every line, then convert it into a cost envelope per department. Issue one template, locked, with the assumptions on the tab they affect.

Consolidate into one workbook with one chart of accounts and one version tag, run the review round as a written explanation exercise, and take the open items to the challenge meeting. Lock version 1.0 after approval, load the phasing into the monthly pack, and forecast separately for the rest of the year.

None of that requires new software. It requires a calendar with dates people respect, a template that asks for the basis rather than the number, and a budget that is worth measuring against in month nine.

Related reading

• Master Budget and Forecast Models
• Budget vs Actual for Not-for-Profits in Excel
• Driver-Based Financial Forecasting in Excel