Bottom-Up vs. Top-Down Forecasting

Two directions

Top-down forecasting begins with a goal — say, '\$2M in revenue next year' — and allocates that across months, channels, and clients based on historical mix. It's fast, makes targets explicit, and is the default for board-level planning.

Bottom-up forecasting begins with a list: every active client, every open opportunity, every recurring engagement. Each is forecast individually with assumed revenue and timing, then summed. It takes longer but reveals concentration risk and timing detail you can act on.

When the two disagree

Mature finance teams build both and reconcile the gap. If top-down says \$2M and bottom-up only finds \$1.4M of identifiable demand, the missing \$600K is either a pipeline-development goal that needs a marketing plan attached or wishful thinking that needs trimming.

For cash forecasting specifically, bottom-up is essential at short horizons because cash timing depends on which client pays when. Top-down works at longer horizons where individual client variance averages out.

Reconciling the two

The most reliable forecasts triangulate. Build a bottom-up model from your real pipeline, headcount plan, and signed contracts. Build a top-down model from market size, win rate, or year-over-year growth assumptions. Then compare them. If the gap is large, you have to explain it: either the pipeline is missing deals you expect to materialize from somewhere, or the top-down number is aspirational.

For early-stage companies the bottom-up model is often laughably small (the pipeline only covers two quarters) and the top-down number is often wildly optimistic (the founder's TAM math). The reconciliation forces a productive conversation: what specific go-to-market motions, in what quarters, would close the gap? That gives the sales and marketing teams an actual plan instead of a target.

For mature businesses the two methods should converge to within ~10%. A persistent gap is a sign the planning process is broken: either sales is sandbagging from below or leadership is force-fitting from above, and the resulting budget will be wrong in predictable ways.

Whichever method dominates, write down the one assumption you would most need to be wrong about. For a top-down model it's usually win rate; for a bottom-up model it's usually pipeline timing. Naming the riskiest assumption explicitly turns 'is this forecast right?' into 'are we tracking what would invalidate it?' — a much more productive ongoing conversation.