Your first forecast

What happens after you connect data, and where to look first.

2 min read

The moment data lands, Foreqast sorts every transaction into canonical categories and projects each one forward, up to 24 months. There is no modelling step to complete first — the forecast exists as soon as the actuals do.

Where to look, in order

1.Dashboard — cash on hand and runway. Check the cash figure against your bank first; everything downstream counts up and down from it.

2.The Foreqast table — every category, with the projection stacked on the booked actual. This is where you find out why a number is what it is.

3.A driver or a contract — the first thing worth changing, once you have seen what the defaults produced.

Sanity-check these three before trusting it

•Revenue and payroll. They carry most of the forecast. If a month you remember well looks right on both, the rest usually is.

•The Other category. Large means transactions are not being classified, so real lines are missing spend. Fix the mapping rather than the forecast.

•Anything periodic. Quarterly tax, annual insurance, yearly SaaS renewals. The default projection spreads or misses these, and they are the most common reason a first forecast looks better than reality. Model them as contracts.

Why the first version looks too smooth

Every line starts on the trailing 30-day average, which by construction produces an even, unremarkable forecast. Real businesses are lumpy. That smoothness is not the forecast being wrong so much as it not yet knowing what you know — which lines are seasonal, which are about to change, which arrive in one payment a quarter.

Changing three or four lines usually accounts for most of the difference. Forecasting models covers what to switch them to.

The forecast updates itself as new actuals import, so this is a one-time set-up rather than a monthly rebuild.

PreviousConnect your data Next Reading the forecast table