Software & SaaS 45 staff, CHF 9M ARR · 5 weeks, then quarterly reviews
A software company's forecast, from ±28% to ±6%
The board had stopped trusting the forecast. A driver-based model built on cohorts, pipeline stages and hiring dates brought the quarter-ahead error down to six percent.
Forecast error, quarter ahead
Before
±28%
After
±6%
The company had grown to CHF 9M of annual recurring revenue on a forecast that was a top-down growth rate applied to last year. It missed by more than a quarter in either direction, three quarters running, and the board had started to discount everything the finance team presented.
What we did
We rebuilt the forecast bottom-up: revenue by customer cohort and renewal date, new business by pipeline stage with conversion rates from two years of CRM data, and costs by named hire and start month. Every driver lives on one tab, and the model reconciles to the accounts each month.
The result
In the four quarters since, forecast error has stayed within six percent. The company used the model to time a hiring plan and a price increase, and the finance lead now runs it without us, with a quarterly review.