Driver-based forecast linking ops decisions to cash.
June: half a year of actuals demands a reforecast, not just a variance.
Refreshes on the 1st of every month
Don't pretend the original plan still holds if the world changed. Reforecast H2 using H1 actuals + revised drivers, then explain the delta vs. plan in writing.
The best models forecast 8-12 drivers — not 400 line items. Pick drivers that someone in the business owns and can move.
| Industry | Top revenue drivers |
|---|---|
| Services | Billable hours × bill rate × utilization × headcount |
| SaaS | New ARR, gross retention, net retention, ARPU |
| Distribution | Active customers × avg order × order frequency |
| Manufacturing | Units sold × ASP × mix |
| Construction / project | Backlog × burn rate × win rate on bids |
The three statements link through specific anchors. If your model doesn't roll forward cleanly, one of these is broken.
Don't build 9 scenarios. Build 3 that tie to real decisions.
| Scenario | Driver delta | Decision it informs |
|---|---|---|
| Base | Plan | Operating budget, hiring plan |
| Downside | Revenue -15%, GM -200 bps | Covenant headroom, hiring freeze trigger |
| Specific upside | New product/market modeled | Capital allocation for that bet |