The Forecast Accuracy section measures how well the system's demand forecasts matched actual order demand.
Note: This feature is currently being beta tested. Check future release notes to see once added to general release.
Average Accuracy
This shows a single percentage which represents how close forecasts were to reality over the last 30 days. 100% means every forecast matched actual demand, lower values indicate worse predictions.
Note: This metric only considers records where the actual demand was more than 0.
Total Forecasts Evaluated
The number of forecast predictions that have been checked against real demand in the last 30 days, based on your current client and warehouse filters. A higher number means the accuracy stats on this page are based on more evidence and are more trustworthy. If this shows 0, there hasn't been enough historical data to evaluate any forecasts yet.
Average Absolute Error (units)
The average number of units by which forecasts were wrong, regardless of over-forecasted, or under-forecasted.
Average Percentage Error
On average, how far off your forecasts were compared to what actually sold, expressed as a percentage.
Actual vs Predicted Demand (Chart)
This is a line chart with two lines, Predicted Demand (purple) and Actual Demand (green), plotted over time for the last 30 days. Where the two lines closely overlap, forecasts were accurate. Large gaps indicate days where the system's predictions were less accurate.
Accuracy Log (Data Table)
This is the drill-down detail. Users can search by SKU or product name, page through results, and toggle column visibility to inspect individual forecast performance records.
What shows on the table:
SKU.
Product Name.
Forecast Date.
Forecasted.
The predicted demand value.
Actual.
The real demand that occurred.
Abs Error.
Accuracy.
Displayed as a colour-coded badge.
Method.
The forecasting algorithm used.
Logged.
When the accuracy log was created.