Driver-based
Headcount drives payroll, pipeline drives revenue, revenue drives commission, headcount drives facilities. Change one input and everything downstream moves rather than needing manual updating.
Platform · financial core
Most forecasting produces a number and then discards it. Next month a new forecast replaces the old one, nobody records which was wrong or by how much, and after three years of forecasting the organisation has learned nothing about its own forecasting.
What it does
Headcount drives payroll, pipeline drives revenue, revenue drives commission, headcount drives facilities. Change one input and everything downstream moves rather than needing manual updating.
An annual budget that stays fixed as the comparison point, and a rolling forecast that updates — because a budget that is revised is no longer a target.
Base, upside, and downside held simultaneously with the driver differences visible, rather than three workbooks that have quietly diverged in other ways too.
What was forecast, when, and by whom — retained permanently, so forecast accuracy becomes a measurable property rather than a matter of recollection.
Budget to actual broken into volume, rate, mix, and timing rather than reported as a single number with a comment box underneath it.
Plan and actual share a chart of accounts and dimensions, so comparison needs no mapping and cannot drift out of alignment.
Almost no company measures how good its forecasts are. Each month a new one replaces the last, the old one is overwritten, and the question of whether the team systematically over-forecasts revenue by nine percent is unanswerable.
Retaining every version makes it answerable, and the answer is usually a stable bias rather than random error. Knowing that a team’s revenue forecast runs consistently high and their cost forecast consistently low is worth more than any improvement to the model.
A line-item budget has hundreds of cells and no structure, so a change in assumption means updating dozens of them and hoping nothing was missed. It also cannot answer the question that matters, which is what would have to be true for this plan to hold.
A driver model states the assumptions explicitly — twelve new engineers, average fully-loaded cost, a start-date curve — and derives the numbers. When the hiring plan slips, payroll, recruitment cost, facilities, and software all move together because they were always linked.
A budget is a commitment made once and held fixed, because its purpose is to be the thing performance is judged against. A rolling forecast is a current best estimate that updates.
Companies that revise the budget mid-year lose the comparison point entirely and usually end up with neither: no target, and a forecast that is treated as a target. Holding both, with the budget locked, is what keeps the variance conversation meaningful.
“Revenue was $180,000 under budget” is not information. Decomposed — volume was under by $340,000, rate was over by $110,000, mix contributed $50,000 — it points at a specific thing to do.
That decomposition is arithmetic and requires the plan and the actuals to share dimensions. It is one of the strongest arguments for planning on the same data rather than in a separate tool that maps across.
Limits
Complex workforce modelling, allocation-heavy planning, and consolidation-grade multi-currency planning at scale belong in Anaplan or Pigment.
A driver model requires the organisation to state its assumptions explicitly, which is uncomfortable and is most of the setup work.
A driver model extrapolates relationships. A new product, a new market, or a shock is a scenario you author rather than something it derives.
Questions
Tell us how you plan today. Retaining versions is what makes accuracy measurable at all.