A forecast that starts with last year's revenue and applies a growth percentage is easy to build and almost impossible to defend. When the board asks why revenue is up twelve percent, the answer is "because we typed twelve percent." When a lender asks what happens if you lose your two biggest customers, the model has no way to answer.

Driver-based forecasting builds the numbers from the operating factors that actually move them. Revenue becomes price multiplied by volume. Wages come from a headcount plan. Materials respond to units produced. The forecast explains itself, and it can be tested. Change one driver and the whole model responds in a way you can trace.

This guide shows how to structure a driver-based forecast in Excel, with an Australian worked example of a small services business. It assumes basic modelling habits. Each assumption lives in its own input cell, named ranges are used where they help, and formulas read from an assumptions sheet rather than carrying numbers inline.

Why Growth Percentages Fail

The flat growth approach has one virtue: it is quick. Take last year's revenue, multiply by 1.10, and the top line appears. The problems show up the moment anyone engages with the number.

First, it hides the mechanism. A ten percent revenue increase could come from ten percent more customers paying the same fees, or from the same number of customers paying ten percent more. Either path produces the same top line, but the two worlds have completely different cost profiles. A services business taking on more clients needs more delivery staff; a business raising prices may not. The growth rate cannot tell you which world you are in.

Second, it ignores cost behaviour. In a percentage-based forecast, wages are usually "last year plus ten percent", which assumes headcount moves in lockstep with revenue. That is true for some businesses and badly wrong for others. A consultancy that adds one senior hire in March to win a pipeline that pays off from June will show a wage spike with no matching revenue for several months. A percentage forecast smooths that timing away and hides the real cash story.

Third, it cannot answer questions. Sensitivity analysis, scenario planning and lender conversations all need to change one thing at a time. Driver-based models are built for exactly that.

The Driver Table: Where the Forecast Starts

Every driver-based model starts with an assumptions sheet. Each driver gets its own cell, labelled clearly, so the forecast schedules can reference it.

A simple driver table for a services business might look like this:

The power of this layout is that every number has a story. "Twenty-eight paying clients" can be defended or challenged. "Three new clients per month" ties to the sales pipeline. The assumptions sheet becomes the document the business argues about, which is exactly where forecasting arguments should happen, before the model is built, not after the numbers come out.

Revenue: Volume Times Price, Client by Client

Revenue in a driver-based model is rarely one formula. It is usually a monthly schedule where each period's client count rolls forward.

Client count follows a simple flow:

closing clients = opening clients + new clients - churn

In Excel, that becomes a schedule. If the assumptions are on a sheet named Drivers, and month 1 opens with 28 clients, month 2's count references month 1's closing balance:

=Drivers!B5 + Drivers!B6 - Drivers!B7

Monthly revenue is then the average client count times the average fee. Using the opening and closing average keeps the timing honest when clients join mid-month:

=((opening clients + closing clients) / 2) * average monthly fee

The schedule shows the shape of the year. Client count climbs steadily as new business outpaces churn, and revenue steps up as the client base grows. If the business signs three larger clients in March, the model captures the step change because the driver changed in March, not because the annual growth rate got tweaked.

This is where driver-based modelling earns its keep in Australia. A flat twelve percent growth line cannot show that revenue dips in January because two contracts end and the replacement work starts in February. The client-flow schedule shows exactly that, because churn and new business are separate drivers.

Wages and On-Costs: The Headcount Plan

The second biggest line in most services businesses is wages, and it is the line percentage forecasts handle worst. Driver-based models build wages from a headcount plan.

The headcount schedule lists each role, the month the person starts, and the salary:

The on-cost rate is where Australian specifics matter. Superannuation guarantee is 11.5 percent of ordinary time earnings from 1 July 2024, rising to 12 percent from 1 July 2025. Payroll tax applies above the threshold in most states, and workers compensation insurance adds a percentage that varies by industry and claims history. A fully loaded cost is rarely the salary itself; 12 to 20 percent on top is the common range depending on the state and role.

Monthly wages per role become:

=IF(month >= start month, annual salary / 12 * (1 + on-cost rate), 0)

The formula charges the fully loaded cost from the month the person starts, and nothing before that month. A January hire appears as a wage step in January. The forecast shows the real timing of the cost, which matters for cash flow and for the lender who wants to see that the hire is funded before the revenue it generates arrives.

Variable Costs: Cost per Unit of Activity

Costs that scale with activity get the same treatment as revenue, in reverse. Materials, subcontractors, delivery and commissions all respond to volume, so they are modelled as a rate times the activity driver.

A business that bills projects and uses subcontractors for overflow work might model subcontractor cost as a rate per project:

=project volume * subcontractor cost per project

A product business models cost of goods as units sold times cost per unit, with the cost per unit reviewed annually for supplier price movements.

The discipline is the same everywhere: no cost line gets a flat growth percentage when an activity driver exists. If the line moves with volume, it gets a rate. If it moves with headcount, it references the headcount schedule. If it is genuinely fixed, like rent or software subscriptions, it stays as a period amount with its own assumption cell, and that is a deliberate choice rather than a default.

Fixed Costs and the Rest of the P&L

Not everything needs a driver. Rent, insurance, software and marketing retainers are period costs. They get assumption cells with their own logic. Rent steps up at lease renewal, and software grows with seats.

The principle is to reserve driver treatment for lines where activity explains the movement, and to be honest about the rest. A forecast where every line is driver-based is over-engineered. A forecast where the big lines are driver-based and the small lines are sensible period amounts is a working management tool.

The completed profit and loss pulls each schedule together. Revenue comes from the client schedule, wages from the headcount plan, variable costs from activity, and fixed costs from the assumptions sheet. Every line traces back to a driver cell, and the model documents itself through its references.

Testing the Model: What-Ifs That Mean Something

A driver-based model's real value is that it can be interrogated. Because each driver is its own cell, a scenario becomes a new set of driver values.

The question "what if we lose our two biggest clients?" stops being a vague worry. It becomes a concrete model change. Reduce opening clients by two, or add two to the churn driver for the month the contracts end. The model shows the revenue dip, the wage underutilisation in the months before staff can be redeployed, and the cash position through the recovery.

"What if the pricing review only lifts fees eight percent instead of ten?" becomes a single cell change on the assumptions sheet.

"What if the January hire is delayed to March?" shifts the wage step and shows the funding relief.

Scenario management in Excel, whether through a scenario selector or simply by saving alternative assumption sets, works precisely because the model is driver-based. A percentage-based model cannot be tested this way, because there is nothing to change that maps to a real business decision.

Australian Worked Example: Putting It Together

A practical example ties the pieces together. Consider a small Australian IT services business with 24 clients and four delivery staff, planning a growth year.

The drivers are set on the assumptions sheet:

The client schedule rolls forward month by month. Revenue is average client count times the fee for that month. Wages show four existing staff from July, and the fifth from January. Subcontractor cost is 8 days times the rate, and it scales if project volume rises.

The forecast shows a business whose revenue grows steadily through the first half, steps up in January when the new business target rises and the new hire lands, and whose margins dip in the middle months while the senior consultant is ramping and billing less than full capacity. That shape is real. A percentage-based forecast would show none of it, which is why driver-based models change the quality of the conversation with lenders, boards and management.

When a Percentage Forecast Is Fine

Driver-based modelling is more work than a growth-rate forecast, and it is not always the right tool.

For a stable business with few moving parts and a forecast horizon of a few months, a simple trend-based projection is often adequate. For a business where revenue is genuinely smooth and costs genuinely fixed, the driver machinery adds little. And for a rough planning conversation rather than a committed forecast, a top-down estimate is a fine starting point.

The test is whether the business has drivers worth modelling. If pricing, volume, headcount or cost structure are changing, or if the forecast will be shown to a lender, board or investor, driver-based logic earns the extra effort. If the business is stable and the forecast is internal, a simpler approach may serve.

Conclusion

Driver-based forecasting replaces the question "what growth rate do we type in?" with the question "what actually drives this business?" Revenue comes from clients and fees, wages come from the headcount plan, and variable costs come from activity. Every number in the model traces to a driver that a manager can defend, challenge or change.

The payoff is a forecast that explains itself, a model that answers what-if questions, and a conversation with lenders and boards grounded in the real mechanics of the business rather than a percentage. For Australian SMEs planning a growth year, that is the difference between a budgeting exercise and a management tool.

Related reading

• Master Budget and Forecast Models for Your Business
• Three-Way Financial Models for Business Decisions
• Cash Flow Forecasting for Growing SMEs