As explained in the article Add or update sales data to SKU Science, you can get new sales forecasts every cycle if you upload your sales data regularly.
This will allow you to track key performance indicators (KPIs) to improve your forecast cycle after cycle, which should be your ultimate goal.
Note that this feature is only available if you have a "Core", "Pro" or "Enterprise" subscription.
1. Viewing the KPIs at the lowest forecast level
You can view the forecast KPIs at the lowest level or at the aggregate level.
Let's start with the lowest level.
If you access the forecast details, from the Navigator tool, you can view the KPIs by clicking on the Key Performance Indicators tab. This tab allows you to view the KPIs at the lowest level only.
Understanding the Lag concept
A drop-down menu allows you to select the lag for which you want to track KPIs.
The lag is the number of periods between the generation of the forecast and the target forecast period.
The lag you select should generally be equal to your lead time. Since lead time is the time after which the item or the material is available, your forecast should be as accurate as possible taking this particular lead time into account.
For example, if you need 3 months between the production of an item and the time for this item to become available to customers, you must anticipate demand 3 months before actual sales. Therefore, you must take care to create an accurate forecast in June for September. In other words, you need to monitor the accuracy of your forecast at Lag 3 and try to improve it. The other lags are less interesting for you because there is a low value to improve your forecast at Lag 1 or Lag 2.
Understanding the values in the KPI table and the forecast value added
Below are the formulas used to fill in the KPI table for each forecast:
A first column displays the average values of the key performance indicators. Each column then displays the KPI for a given period, based on the selected lag. For example, if you display Lag 3, the KPIs in the April column are calculated by comparing the forecast you generated in January (for April) to the actual sales in April.
There are 4 different KPIs:
Bias: the bias is the difference between the forecast and demand (f-d), in units.
Bias (%): ratio between bias and demand, in percentage (100x(f-d)/d).
Error: the forecast error is the ratio of bias to demand, in absolute values. Note that the Average is the mean absolute error (MAE).
Accuracy: (100 - error), in percent.
For each KPI, there are 3 lines:
SKU Science: KPI for forecasts generated by SKU Science.
User: KPI calculated taking into account your adjustments.
Added value: improvement (or degradation) due to your adjustments. A negative added value is an improvement in the case of biases and errors. A positive added value is an improvement in the case of accuracy.
The colors of the cells allow you to compare the KPI values to the thresholds defined by the user:
Red: the error or bias is greater than the upper threshold.
Orange: the error or bias is between the lower and upper thresholds.
Green: the error or bias is smaller than the lower threshold.
The thresholds can be modified in Parameters / Rules / KPI Thresholds.
2. Viewing KPIs for all forecast levels in the dedicated tool
To view the KPIs at different levels of aggregation, click on %KPI in the left-hand menu. You can display all KPIs for the element you are interested in by clicking on the details icon in the table.
You will then get the same type of table, but at an aggregated level, for example, here we see the KPIs on quantities for the London warehouse.
In the KPI tool, however, it is easier to use the drop-down menu to choose which KPI to display in the table.
The forecast error
For example, by selecting Error in the drop-down menu, you will get the error table for all your elements. Note that a Total row is available at the bottom of the table to instantly view the sum of the errors for each column.
The percentage of forecast error
You can also view the error percentage instantly by choosing Error %.
The forecast bias
By selecting Bias from the menu, we display the bias and the sum of the biases.
The percentage of forecast bias
By choosing Bias % it is also possible to display the percentage of bias.
The percentage of forecast accuracy
Finally, if you prefer to see the accuracy rather than the error, you can also select this option from the drop-down menu, and get the following table:
You know how to easily switch from one KPI to another for a given lag. If you have any questions, don't hesitate to contact us directly in the tool's interface, using the messenger at the bottom right of the screen.