← All free tools

Free DTC diagnostic

Forecast Confidence Check

A forecast that's consistently wrong in the same direction isn't noise — it's a broken model. Enter the last 6 months of forecast vs. actual revenue and we'll tell you your accuracy, where the bias lives, and how wide your confidence intervals should actually be.

Enter the last 6 months of forecast vs. actual

Enter your forecasted revenue and actual revenue for each of the last 6 months. Use the most recent month first.

Month
Forecasted Revenue ($)
Actual Revenue ($)
Variance
Optional: dimension weighting

How this works

What is MAPE and what's a good score?

MAPE (Mean Absolute Percentage Error) is the average absolute deviation of your forecasts from actual results. Under 10% is strong for monthly revenue forecasting. 10–20% is normal for most DTC brands with reasonable models. Above 25% suggests the forecasting inputs, model structure, or assumptions need structural work. See our guide on driver-based ecommerce forecasting.

What is forecast bias and why is it worse than random error?

Forecast bias is a systematic tendency to forecast too high or too low. Bias compounds: a brand that consistently overforecasts by 15% will over-plan inventory, over-hire, and over-commit on fixed costs — then scramble to cut when actuals land below plan. Random error around a correct mean is manageable. Systematic bias in one direction creates structural planning problems.

How should I use the confidence interval suggestion?

The suggested confidence interval is calibrated to your historical MAPE. If your MAPE is 18%, a defensible next-month forecast range is roughly ±20–22%. Present your forecasts as ranges in planning documents, not point estimates. A range of "$320K–$390K" is more useful than "$355K" because it forces the organization to plan for both the high and low case.

What data should I enter for "forecasted revenue"?

Enter the revenue number your team formally committed to at the start of each month — ideally from a planning doc, board report, or weekly review. If your process is informal, use whatever top-line projection was circulated before the month started. Don't retrofit better numbers after you saw how the month landed.