XYZ analysis overview
The XYZ analysis is about determining the forecastability or predictability of customers or items, it depends what we want to segment. This is an important factor to consider when going for a granular level forecasting or analysis and this kind of highlight how difficult it is to forecast certain customers or items. XYZ analysis is also about splitting the entire group into 3 main categories, X,Y and Z, where:
- X represents the most forecastable group and the items are demanded frequently, perhaps daily. With frequent replenishment, they should have low demand variation that is predictable.
- Y represents, moderately forecastable group and items are seeked less frequently, such as every few weeks or less. There is more variation in the demand amount.
- Z represents the least forecastable group, items requests are infrequent and irregular. They have the most demand variation and are the most difficult to predict.
XYZ analysis details
The next question that arises is how do we calculate forecastability and rank items based on forecastability? Thre are two approaches:
- The most suggested approach everywhere is calculating Coefficient of Variation (CoV) which is essentially calculation a value of standard deviation divided by mean (scaled standard deviation). But from practical experience, we have often seen many flaws with this approach. This method only looks at the historical variation but does not consider seasonality and is susceptible to outliers.
- From experience, a better approach is to consider forecast accuracy, mostly by using Mean Absolute Percentage Error (MAPE) between the actuals and the forecasted values. So, the idea will be to consider about 80% of the historical time series data to predict the remaining 20% time series data and compare the predicted with the actuals to calculate MAPE and then rank all the items by the predictive model MAPE values and use a similar splitting ratio of top 20% of X cluster, next 30% for Y cluster and last 50% for Z cluster.
Domains of implications
XYZ analysis (or XYZ classification) can be used in various domains and for various item segmentations. We can highlight some of them :
- Retail – Demand planning are directed towards ensuring maximum revenue and maximum utilization of available resources
- By SaaS product teams for segmenting their users based on their platform activity, so here the users are split based on the frequency of visit and platform usage, are they frequent or not so frequent respective X, Y or Z segment
- Manufacturing – Supply chain management strategies are focused to ensure maximum utilization of available resources and focused ability to convert big opportunities to bigger
For one of our customers they have SaaS product and are interested to measure the frequency of usage of their users, so XYZ gave them information about which customers are more frequent than the others and also if you combine them with ABC analysis which are volume based then there are 9 segments which can be used from the marketing teams for personalized messages and approaches.
Catwing helps every online business to benefit from XYZ Analysis
Catwing is a flexible solution which offers XYZ analysis tailored to various businesses and XYZ analysis is part of the basic plan. Our clients can execute full XYZ analysis based on user-defined criteria, risk analysis and schedule optimization. Learn more about other models and our pricing plans.