ABC/XYZ Segmentation- Examples
Here is an example to demonstrate the various ABC/XYZ segmentation methods available in IBP. While there is a lot of information already available on ABC/XYZ, I am sharing an example to hopefully help you understand it better. I would still recommend reading through SAP IBP help to get the high-level details before navigating through the following example.
Given below is the Actuals Qty. data for the last 6 months for 8 unique products FG1 to FG8. In the column at the far right, is the sum of Actuals Qty. for those 6 months. We are going to do segmentation based on this data. Now, this doesn’t mean that you must always take Actuals Qty. as the segmentation measure (source Key figure). Nor does it mean that you must always take data for the past 6 months. These would vary from business to business. The purpose of this blog is to just explain the technicality of ABC/XYZ methods using this simple example.
Before moving further, lets understand the different fields within the ABC segmentation profile. Fields shown below (Fig 01) have the exact same meaning even for XYZ segmentation profile.
We will now see the outcome of all ABC segmentation methods. Let’s start with the two methods associated to Pareto principle:
Significance of grouping: “Group” is basically the level at which recalculation of ABC segmentation starts. For example, so far, all the 8 test products FG1 to FG8 were being considered together for segmentation. When we use grouping, the products are segregated in separate groups based on the grouping parameter and segments are calculated individually for each group. Products in one group have no influence on the segmentation results for products of the other group.
For example, we are going to perform segmentation based on Product Family(Fig 02). Our test products can be divided into 2 Product Family groups- White Chocolate & Dark Chocolate. Based on the outcome below you will notice that now, ABC segmentation is done separately for White Chocolate (FG1-FG4) v. Dark Chocolate (FG5-FG8) product families.
Moving on to XYZ segmentation:
Calculation strategies for XYZ Segmentation:
There are basically 2 calculation strategies for XYZ segmentation:
1. Calculate Variation– With this strategy, variation of segmentation measure (Actuals Qty. in our case) is calculated, and products are segmented in X, Y & Z segments based on the comparison of variation with the thresholds defined. There are 2 Calculation methods associated with this strategy:
- Coefficient of Variation(CV)– With this strategy, the thresholds are compared against Coefficient of variation (CV= Std. Dev/ Mean).
- Coefficient of Variation(CV) squared– With this strategy, the thresholds are compared against Coefficient of variation squared.
See below the outcome of XYZ segmentation for Calculation Strategy= Calculate variation for both calculation methods side by side for the thresholds given below:
2. Aggregate over periods– Another strategy for XYZ segmentation is “Aggregate over periods”. Under this strategy, the segmentation measure is aggregated using the selected aggregation method (either of Minimum/Maximum/Average/Sum) and compared with thresholds to identify segments.
Let’s say, you want products to be segmented based on MAPE error values that were calculated using the “Manage Forecast Error calculations” app. This is how you can model it:
*K Means- This segmentation method can be used when you have a large data set, and you don’t know what thresholds should be set for different segments. K Means is a method that uses machine learning to identify thresholds for ABC or XYZ segments and then distribute the product in those segments accordingly.
I hope the example shared above clarified ABC/XYZ segmentation better. You can read other posts relevant to IBP Demand at these links:
Here is another blog on ABC/XYZ segmentation to understand the behavior of the system when the cumulative percentages do not exactly add up to the threshold boundaries.