A RANGE MODEL LEARNING TOOL
Backtest the Baseline
Go back in time. Hide the future. Make the forecast. Then reveal what actually happened.
Use only what you could have known then.GUIDED EXERCISE01-03 / FORECAST
STEP 1 / GO BACK
It is January 2025.
You are forecasting organic-search sales for January-December 2025 without any new SEO work.
THE RULEYou may only use information available up to this date.
STEP 2
What you know
Fictional teaching data - not an industry benchmark- Previous 12 months
- 920 sales from organic search
- Recent trend
- +4% in the latest six months versus the previous six
- Seasonality
- Q4 is usually stronger than Q1
- Previous SEO work
- A major category-page update launched five months ago
62Jan
64Feb
68Mar
70Apr
73May
75Jun
78Jul
79Aug
81Sep
88Oct
91Nov
91Dec
STEP 3
Your forecast
What happens without the new SEO work?No peeking. The restriction is the point.
OPTIONAL / YOUR OWN NUMBERS
Backtest one of your own old forecasts.
Nothing entered here is sent or stored. Keep commercially sensitive notes out if you are sharing your screen.