Imagine the Vice President of Sales enters your office and explains that he just promised the CEO that he will increase sales from the current customer base. He pauses, then asks for your help. You reflect and realize that the next marketing campaign would be the best place to start. But where do you begin? Your objective; increase the ROI on your next marketing campaign.
You need to decide on your approach and have several options that come to mind:
- Take a shotgun approach where you blanket the customer base with a general message. Last time you tried this, it had a low conversion rate and a high cost.
- Go with an intuitive approach where you review the customer list and use your “gut feel” to identify the target customers. Last time, this provided some good insights but overall was expensive and unreliable.
- Take a traditional statistical approach with regression/classification combined with cluster analysis / segmentation. This provided excellent results but it took longer than expected and was difficult to execute.
- Use SAP Predictive Analytics (PA) using a regression/classification model combined with cluster analysis/segmentation. This approach was relatively faster to use and provided excellent results.
You decide on option four, using SAP Predictive Analytics to get the best ROI from your marketing campaign. You already know that you can use PA to help you predict customer churn, predict consumer buying behavior, and identify upsell and cross sell strategies, so this will be a piece of cake. Now it’s time to review your available data. Predictive Analytics will assist in preparation and data cleansing, eliminating redundant, highly correlated data and selecting a strategy for blank / null values.
You layout the problem to be a two-step process. First you want to identify the customers that are most receptive to a promotion based on their history. This can be done using PA’s Regression/Classification model. You can stop here and have one promotion with your target customers, which increases your yield significantly, but you think you can improve your results further but having separate customized promotions for groups of customers based on their attributes. This can be accomplished with PA’s Segmentation/Cluster Analysis.
To identify the target customers, start with Predictive Analytics and Regression/Classification models. For data, you need to identify the target variable and model sample size. Predictive Analytics will rank the variables and use only relevant data. (you can set the tolerances) Predictive Analytics will generate the model and provide statistics for fit such as Predictive Power and Prediction Confidence scores. This step identifies customers who are best suited for the promotion.
The Segmentation/Cluster Analysis with Predictive Analytics will follow a similar process. Identify the target variable in the data and Predictive Analytics will identify the best fields and generate a model with results based on the test data. That model can be used against the complete data set. This step provides clusters or groups of optimal customers for the specific promotions.
At each step of the process, Predictive Analytics provides statistical evaluation of your data to help you understand your results and the model generated. For example, Predictive Analytics provides the contribution by variable, Confusion Matrix, Decision Tree, Coefficient values and other statistical analysis.
SAP Predictive Analytics Results
In this example, SAP Predictive Analytics (PA) provides a robust solution with much of the detailed work addressed by the automated features of the software. The time savings enables you to work on more beneficial projects. By combining two groups of algorithms, you were able to identify the optimal customers and refine your campaign for the specific market groups, to further increase your campaign ROI. As you see from the above example, applying SAP Automated Predictive Analytics will increase the ROI on the next marketing campaign.