Adding Business Expertise to Sales Forecasts

SkuBrain now lets you explore and tweak your sales forecasts. Once SkuBrain has built a baseline statistical forecast for the products you sell, you can dig into the data and make any adjustments, using the smarts that SkuBrain doesn’t have (like knowledge about when stock-outs occurred or maybe promotions you have planned).

Forecast explorer

The new forecast table, below the chart, shows raw data for the baseline statistical forecast generated by SkuBrain and lets you manage any manual adjustments. By default, to keep things clean and simple, that’s all that will be displayed in this panel but if you’re passionate about statistics (who isn’t right?) then you can get more detail by flipping the nerd nerd icon switch at the top right of the page. This will reveal some extra detail such as the statistical model that was used to generate the forecast and the 80th and 95th percentile confidence intervals, if these are available.

Forecast model with prediction intervals

If looking is fine but you really want to touch then click on the Adjust button underneath the forecast data to pop open the Adjust forecast dialog. In addition to letting you punch in new forecasts manually, here you can run some basic formulas in order to easily calculate adjusted forecast figures for one or more sales periods. The formulas allow you to apply percentage increases to the baseline forecasts, fixed/additive increases or simply to set the forecasts for a bunch of periods to an explicit value.

Forecast adjustment

Last, but not least, you can view the adjustment history for any node in the forecasting hierarchy to see what changes were made when and by who, including specific details of how things looked before and after each change (and yes, rolling back changes is in the works so this will be available shortly).

Forecast adjustment history


It’s worth mentioning that adjustments you make will affect multiple forecasts. So, for example, if you increase the forecast for November for a particular SKU by, say, 10 units, that change will be bubbled up to all of the parent nodes for that SKU in your forecast hierarchy. Implicitly then, the root level forecast (the one labelled All sales in the forecast hierarchy tree) will also be increase by 10 units in November.

Similarly, any adjustments made to forecasts at the root level will be disaggregated down to lower levels of the hierarchy as appropriate. SkuBrain takes care of the heavy lifting though, so you don’t have to worry about the details.

Anyway, that’s it for this week – enjoy!