The summary.
THE SIMPLEST METHOD THAT WORKS
Last year's same period, adjusted for known changes.
For a seasonal business, frequently the best available. Produce a range, not a single figure.
THE FORECAST THAT MATTERS MOST
Cash, by week, over roughly thirteen weeks.
Use likely payment dates, not invoice dates. Late payers pay late in the forecast too.
It reveals the week the balance goes negative, which is what lets you act early.
BUILD THREE SCENARIOS
Expected, better, worse — with the worse case plausible rather than catastrophic.
Decide in advance what you would do, and what the trigger is. Decisions made under pressure are worse.
RECORD YOUR FORECASTS
Compare against actual. Your typical error becomes your range, and a consistent bias can be corrected.
That is the only way forecasting improves.
FOR PRICING
Model the volume you could afford to lose and still be better off.
A modest rise usually survives a substantial volume loss — and quantify what habitual discounting costs annually.
FOR STAFFING
Staff to the peak, not the average. Consistent overtime is a hiring decision deferred; price it annually.
