The annual budgeting process is the six-to-eight-week cycle in which a company turns next year's strategy into numbers people have agreed to be measured against. For most startups and SMBs on a calendar fiscal year it should start in September, run top-down and bottom-up at the same time, and finish with an approved plan before the year begins. Done properly it costs about six weeks of part-time effort from a handful of people; done badly it swallows the fourth quarter and produces a spreadsheet nobody opens again in February.
This is the process I run with clients. It assumes a company between roughly €2M and €20M of revenue, a founder or CEO who wants the plan to mean something, and no dedicated FP&A team.
When should you start?
Work backwards from the date the board approves the plan.
If the board meets in early December and you want the budget approved there, the pack goes out in late November, which means the numbers must be stable by mid-November, which means department inputs land by the end of October, which means you kick off in the last week of September. That is the whole logic. Six to eight weeks, not three.
Most companies start too late — mid-November, in a panic, with the year already gone — and end up doing top-down maths in a weekend. The plan then arrives as a target handed down rather than a commitment made, and the difference shows up all year in how people talk about it.
Two adjustments worth making:
- Non-calendar fiscal year: shift everything, but keep the same shape. Kick-off is roughly three months before year end.
- Fundraising in the window: if you expect to raise in the first half of next year, the budget and the fundraising model are the same object built twice. Build the budget first, honestly, then build the investor case on top of it. Investors read the difference between the two, and they should — it tells them what you actually believe. Our guide on how to build a 3-statement financial model covers the mechanics.
Who owns which part of the budget?
The most common failure in SMB budgeting is not analytical. It is that one person — usually the founder, sometimes the accountant — builds the whole thing alone, and so nobody else feels bound by it.
A workable division for a company of this size:
| Role | Owns | Does not own |
|---|---|---|
| CEO / founder | The strategy the budget expresses; final trade-offs | Building the model |
| Finance (or fractional CFO) | The model, the drivers, consolidation, the calendar | The commercial assumptions |
| Sales / commercial lead | Pipeline, win rates, new-business targets, quota capacity | Marketing spend levels |
| Marketing lead | Spend by channel and the leads it produces | Revenue targets |
| Product / engineering lead | Headcount, contractors, tooling, infrastructure | Hiring approvals |
| Operations / delivery | Delivery cost, utilisation, capacity constraints | Pricing |
The rule: whoever will be asked about a variance in June should have signed the number in November. If your sales lead did not build the new-business plan, do not expect them to defend it.
What does a six-week calendar look like?
| Week | What happens | Output |
|---|---|---|
| 1 | CEO sets strategic frame: growth ambition, profitability or runway constraint, one or two big bets | A one-page brief, not a number |
| 2 | Finance builds the top-down envelope and the driver skeleton; last year's actuals cleaned and restated | A model with drivers empty |
| 3 | Department heads fill in their drivers and headcount asks | Bottom-up inputs |
| 4 | Consolidation. The first version never balances — this is normal | The gap between top-down and bottom-up, quantified |
| 5 | Trade-off sessions. Cut, sequence, or fund the gap | An agreed plan |
| 6 | Scenario overlay, board pack, approval | Approved budget + phasing |
Week four is the point of the exercise. The bottom-up ask always exceeds the top-down envelope — in my experience by something like 15–30% of opex on a first pass. That gap is the conversation you are trying to have. If it does not appear, either the top-down was too generous or the department heads were too cautious, and both are worth knowing.
Top-down or bottom-up? Both, deliberately
Top-down alone gives you a plan that is coherent but fictional: revenue grows 40% because the CEO wants 40%, and nobody has asked whether the sales team has the capacity to deliver it.
Bottom-up alone gives you a plan that is detailed but unaffordable: every department asks for what it needs to do its job well, and the sum is a company that burns twice its cash.
Run them in parallel and force them to meet.
The top-down envelope answers: what can we afford? Start from cash. If you have €4.2M in the bank and the board wants 18 months of runway at the end of next year, the maximum net burn for the year is roughly the amount that leaves 18 months of the following year's run-rate intact. That is a hard ceiling, and it should be set before anyone opens a spreadsheet. (Figures here are illustrative.)
The bottom-up build answers: what does each team need to hit the plan? Headcount by role and start month, tooling per head, spend per channel, delivery cost per unit.
Where they meet is where the real decisions live: which two of the five hires happen in Q1 rather than Q3, whether the new market entry is funded or deferred, whether pricing has to move.
Budget the drivers, not the line items
The single biggest quality difference between a budget that survives and one that does not: whether the numbers are calculated or typed.
A typed budget says marketing spend is €45,000 a month. A driver-based budget says marketing spend is whatever it takes to generate the leads that support the new-business target, at the blended cost per lead we saw last year plus a degradation factor. When the target moves in March, the typed budget is wrong and the driver-based one re-solves.
The drivers worth building for most companies:
- Revenue: for SaaS, opening ARR, new bookings, expansion, churn — a bridge, not a growth rate. For services, billable headcount × utilisation × rate. For e-commerce, sessions × conversion × average order value, with returns netted.
- Cost of revenue: the handful of unit costs that actually vary — hosting, fulfilment, payment fees, delivery salaries.
- People: a headcount plan by role and start month. This is usually 60–75% of opex and deserves its own tab, with fully loaded cost per role including employer taxes, which differ enormously between an Austrian, Hungarian, or US hire.
- Everything else: per-head costs (software, equipment, travel) and fixed costs (rent, insurance, audit) budgeted separately. Do not put them in one bucket.
Three or four well-chosen drivers per department is plenty. A model with forty assumptions is not more accurate; it is just harder to argue with, which is the opposite of what you want.
Phase it monthly, or it isn't a budget
An annual number divided by twelve is not a plan. It is an average that will be wrong in every single month.
Phase revenue for seasonality you can actually evidence — summer slowdown in European B2B, Q4 concentration in e-commerce, the January restart in professional services. Phase costs for when hires start, when the annual insurance premium lands, when the audit fee hits, when VAT and corporate tax are paid.
The monthly phasing is what makes the budget usable for variance analysis and for cash. Without it, your first quarterly review is a debate about whether you are behind or just early in the year, and nobody can settle it. See budget vs. forecast vs. actuals for how the three should relate once the year starts.
Add two scenarios, not seven
Before approval, run the plan at two variants beyond the base case:
- Downside: revenue lands materially below plan — say 25% below — and you take no action for one quarter. Where is cash at year end? What is the runway? This is the scenario that tells you whether the plan is survivable, and it is the one boards ask about first.
- Upside: the plan works and you want to lean in. What would you spend the extra on, and what would you need to see before spending it?
Attach trigger points to each: if new bookings are below X for two consecutive months, we defer the Q3 hires. A scenario without a trigger is a slide. A scenario with a trigger is a decision made calmly in November instead of anxiously in July.
What breaks the process
Chasing precision. Forecasting Q4 revenue to the euro in September is a waste of a good week. Be roughly right about the drivers and precise about the constraints.
Negotiating with sandbaggers. If department heads learn that asks get cut by 20%, they will inflate by 25%. Break the cycle by asking for a prioritised list rather than a single number: what you would do with the base allocation, and the next three things you would fund in order.
Treating the budget as immutable. It is a snapshot of what you believed in November. Plan to reforecast at least quarterly — this is the whole argument in rolling forecast vs. static budget. Keep the original budget frozen for accountability, and run the forecast alongside it.
Letting the model live on one laptop. Version control, an assumptions tab, and no hardcoded numbers inside formulas. Somebody other than the builder should be able to open it in April and follow it.
FAQ
How long should the annual budgeting process take?
Six to eight weeks of elapsed time for a company under €20M of revenue, which is perhaps 10–15 working days of actual finance effort plus a few hours from each department head. Longer than that and it starts competing with running the business.
Should a pre-revenue or very early-stage startup do an annual budget?
Yes, but a smaller one. Below roughly €1M of revenue the useful artefact is a headcount plan, a spend plan, and a runway calculation — not a full three-statement build. The discipline of naming what you will spend and what you expect back is what matters at that stage. Startup runway management covers the shape of that.
What if we don't have clean historical data to budget from?
Budget from drivers rather than from history. You may not have three years of comparable P&L, but you know your current pipeline conversion, your current cost per hire, and your current unit costs. Use the last two or three quarters as the base and say explicitly which assumptions are estimates.
Who approves the budget — the board or the CEO?
The board approves it where one exists, usually at the last meeting of the year, and the approved version becomes the basis for the following year's variance reporting. The CEO owns it either way. A budget the board has seen but not agreed to will be relitigated at every meeting.
Where to start
If your last budget process ended with a number nobody remembers, the fix is usually structural rather than analytical: start earlier, share ownership, budget drivers instead of line items, and phase the result monthly.
If you want an outside read on whether your planning process is doing its job, the free financial diagnostic takes five minutes and scores your finance function across planning, reporting, controls, and unit economics. If you would rather talk it through before budget season starts in earnest, get in touch — a 30-minute call is usually enough to work out whether the process needs a tune-up or a rebuild.