A rolling cost forecast maintains a continuous 12 or 18 month view of spend, extending by one period as each period closes. It suits businesses whose cost base moves with hiring, and it depends entirely on a close that lands fast enough for the update to still be a decision.
What rolling cost forecasting is
An annual budget is set once and is at its least useful in the month before it is replaced. A rolling forecast fixes the horizon rather than the end date: when March closes, March drops off the front and a new month is added at the back, so the window stays the same length all year. Applied to costs specifically, it means the hiring plan, contractor spend, tooling and overheads are re-forecast every period against what actually landed.
The case for it is strongest in businesses where the cost base moves with headcount, which is most Australian scale-ups. For founders and CFOs scaling fast, often with external funding and burn rate front of mind every month, what is needed is finance people who can do the work and build the process while staying lean.[1] A rolling forecast is one of those processes, and it does not build itself.
How to set the window up
Start from history rather than a blank sheet. Take the last twelve months of actual cost by category, strip the one-offs out into their own line so they do not get baked into the baseline, then project each category forward on a stated driver: headcount for payroll, seats for software, revenue for anything variable. Payroll is the one people under-project, because a head costs the package plus the on-costs sitting on top of it, starting with the superannuation guarantee the ATO requires employers to pay. Twelve months is the usual window. Eighteen is worth it if your hiring decisions have long lead times.
| Annual budget | Rolling cost forecast | |
|---|---|---|
| Horizon | Fixed end date, shrinking all year | Fixed length. You are always looking the same distance ahead, which is what makes the number comparable month to month. |
| Update cycle | Once a year, plus a re-forecast when things go wrong | Every period, as part of close. The update is small because it is frequent. |
| What it is used for | Accountability against a committed number | Decisions. Mainly whether the next hire, tool or contract fits before it is signed. |
| Main failure mode | It stops describing the business by about month four | It becomes a monthly ritual nobody reads. The tell is budget holders returning last period's numbers unchanged. |
Then decide, before you start, what the update is for. If the monthly re-forecast does not change a decision, it becomes an admin ritual within two quarters and people stop reading it. The version that survives is the one where the cost forecast is what tests the hiring plan before a role is signed off.
The rolling forecast reads from the closed actuals, so it is only as fast as whoever owns the profit and loss and the month-end close.
What it costs to run
The benefits are real: the plan never goes stale, the year-end budgeting marathon shrinks to a review, and cost decisions get made against a live number instead of a nine month old one. The costs are equally real and less discussed.
The first is close speed. A monthly re-forecast is worthless if the actuals arrive three weeks after the period ends, because by then the decision has already been made without them. It is an achievable target, and I have worked with a candidate who cut a monthly close from 28 days to 7 within two reporting cycles.[2] That is the prerequisite, not a nice-to-have. The second cost is fatigue, on both sides. Finance gets tired of the cycle and budget holders get tired of being asked the same questions twelve times a year, so the discipline has to be worth something visible to them.
Tools, and where to start
Most Australian companies at this size run this in Excel or Google Sheets and that is a perfectly good answer, which is also the consensus among the startup finance specialists in our panel on founder built financial models. Dedicated FP&A tools earn their keep when the data pulls are the bottleneck rather than the modelling, and an ERP will only help if the cost data in it is already clean.
On automation, the sequence matters more than the tool. My view is to document how the finance process truly works first, then use AI to identify the risks and the timesinks, then automate one small step at a time.[3] Automating a rolling forecast on top of a process nobody has written down produces a faster wrong answer.
Who owns it, and what that means for hiring
A rolling forecast fails without a named owner, because it is a recurring commitment rather than a project. In a company at this stage that owner is usually the first proper finance hire rather than the founder or the bookkeeper. In the Australian market a first finance hire in the $140k to $160k range means a Finance Manager or Financial Controller who can own the function, build the processes, manage the external advisors and give the business visibility.[4] That last word is the whole job description for this work.
The interview question worth asking is not whether a candidate has run a rolling forecast. It is what they changed when the forecast and the actuals disagreed, and whether the budget holders were still talking to them by the end of the year. Rolling forecasting is a stakeholder discipline dressed as a modelling one.
If nobody obvious owns it today, the underlying question is how a startup finance team should be structured as it scales.
Common questions
What is rolling costs forecasting?
It is a method that keeps a fixed forward window of cost forecast open, usually 12 or 18 months, and extends it by one period each time the current period closes. Instead of an annual budget that ages all year, you always have the same distance of visibility ahead. Applied to costs it means the hiring plan, contractor spend, tooling and overheads are re-forecast every period against what actually landed.
How do you build a rolling cost forecast?
Start from history. Take the last twelve months of actual cost by category, separate the one-offs so they are not baked into the baseline, then project each category forward on a stated driver: headcount for payroll, seats for software, revenue for anything variable. Twelve months is the usual window, eighteen if your hiring decisions have long lead times. Decide up front which decision the monthly update is meant to inform.
Is a rolling forecast better than an annual budget?
It is better at decisions and worse at accountability, so most companies end up running both. The budget stays as the committed number, and the rolling forecast is what actually gets used to test whether the next hire or contract fits. It only works if the month-end close lands fast enough that the update can still change something.
What tools do you need for rolling cost forecasting?
Less than vendors suggest. Excel or Google Sheets handles it well at startup and scale-up size, which is also the consensus among startup finance specialists. Dedicated FP&A tools earn their keep when the data pulls are the bottleneck rather than the modelling, and an ERP only helps if the cost data in it is already clean. Document how the process actually works before automating any of it.
References
- Something I notice again and again: for founders and CFOs scaling fast, often with external funding and with burn rate front of mind every month, the need is for finance hires who can do the work and build the process, all while staying lean. Tom Hunter hosts The CFO Track, a podcast of interviews with Australian CFOs and finance leaders.
- A candidate Tom Hunter has worked with cut the monthly close process from 28 days to 7 days within two reporting cycles in their previous role.
- What I see AI actually changing: document how a finance process truly works, then use AI to identify the risks and timesinks, then automate one small step at a time.
- Where the Australian market sits on this: a salary in the $140k to $160k range signifies hiring a Finance Manager or Financial Controller who can own the function, build processes, manage external advisors and provide visibility.