Tools · Forecast Accuracy Scorecard
How good areyour forecasts, really?
Twelve months of forecast versus actual, typed or pasted from Excel. Out comes a score, your accuracy and bias, where the misses cluster, and the range your own history says you should have been quoting all along.
YOUR NUMBERS STAY IN YOUR BROWSER UNLESS YOU ASK US TO EMAIL THE SCORECARD
Your twelve months
0 / 12At least 6 months to score. Nothing leaves your browser unless you ask us to email the report.
YOUR SCORECARD
Type or paste at least 6 months of forecast and actual, or load the example, and the scorecard builds here: accuracy, bias, where the misses cluster and the range your history says you should have been quoting.
Take it with you
What the scorecard measures
A missed budget is data. Most businesses throw it away.
Every month of forecast versus actual tells you something about how the business thinks about its own future: which way it leans, when it gets surprised and how wide its honest range really is. The scorecard turns that history into the two things a board needs: a corrected midpoint and a range it can plan against.
Accuracy · MAPE
Mean absolute percentage error: on average, how far the forecast landed from the actual, ignoring direction. The headline number, but only half the story.
Bias · the lean
The average signed error. If it is consistently positive, the forecasts are optimistic; the misses are a habit, not bad luck. Bias is the cheapest thing to fix and the most expensive to ignore.
Clustering
Whether the misses bunch in particular months or run in the same direction for months on end. Clustering points at phasing and seasonality; runs point at a forecast that is never re-based.
The range · P10–P90
What your own history says a realistic range around each forecast would have been, and how many months it would have caught. It is the range you should have taken to the board.
Related: Probability Forecaster turns next year's ranges into odds. · All tools · Training and resources