The limitation of line-item forecasts
Traditional forecasting takes each line item on the P&L, looks at the trend, and projects it forward. Revenue grew 3% per month, so next month it will grow 3% again. Marketing spend averaged £45,000, so budget £45,000 going forward.
This approach works until it does not. When something fundamental changes -- a new product launch, a pricing change, a shift in customer mix -- trend-based forecasts break because they model the output without understanding the input.
What driver-based planning looks like
Driver-based planning builds the forecast from the operational levers that generate financial outcomes. Instead of forecasting revenue directly, you forecast the drivers: number of leads, conversion rate, average contract value, and expansion rate. Revenue becomes a calculated output.
The same logic applies to costs. Instead of forecasting "travel expense = £30,000," you forecast "15 customer visits at £2,000 each." When the number of customer visits changes, the cost forecast updates automatically.
Identifying your drivers
The 80/20 rule applies heavily here. A handful of drivers explain the vast majority of financial outcomes. For a typical SaaS business:
Revenue drivers: New customer count, average deal size, net revenue retention, pricing uplift, payment terms.
Cost of goods drivers: Hosting cost per customer, support headcount per thousand customers, third-party API costs per transaction.
OpEx drivers: Headcount by department, average fully loaded cost per employee, marketing spend as a percentage of target revenue, office cost per seat.
Start with the five to ten drivers that explain 80% of your P&L. Add granularity only where the incremental accuracy justifies the additional complexity.
Building the driver model
A well-structured driver model has three layers:
Input layer. The assumptions you actively manage: growth rates, conversion rates, pricing, headcount plans. These are the cells that change when you update the forecast.
Calculation layer. The formulas that translate drivers into financial outcomes. Revenue = customers x average revenue per customer. Salary cost = headcount x average salary x (1 + employer NI rate + pension rate).
Output layer. The P&L, balance sheet, and cash flow statement. These are entirely calculated -- nobody edits them directly.
The benefits
Better scenario analysis. When the model is driver-based, scenario planning becomes intuitive. "What if conversion rates drop by 2 points?" is a single input change that cascades through the entire model.
Faster reforecasting. Updating a driver-based model requires changing a handful of assumptions, not reworking hundreds of line items.
Clearer communication. When the CFO presents a forecast to the board, the conversation centres on operational metrics everyone understands -- customer growth, retention, headcount -- rather than abstract financial line items.
The transition
You do not need to convert your entire model to driver-based planning overnight. Start with revenue -- it is the account with the highest impact and usually has the most identifiable drivers. Once the revenue model is driver-based, extend the approach to cost of goods, then operating expenses. Within two to three budget cycles, you will have a fundamentally better planning framework.