Campaign Optimization and SAP NetWeaver BI
SAP NetWeaver Business Intelligence (SAP NetWeaver BI) is used for campaign optimization and to create response predictions models. During campaign execution, the system also saves the predictions produced by the optimization in a DataStore object. If required, you can use queries to analyze these predictions.
No data is saved in the DataStore object when the simulation is being performed.
Features
DataStore Object Direct Marketing Optimization
When you execute the optimization for a campaign, the system saves in the DataStore object Direct Marketing Optimization
(0CRM_ODMO) the detailed assessments for each individual business partner. Decision-making regarding optimization then uses this data as its
basis.
If you would like to analyze the data, you define a query in the InfoSet Direct Marketing Optimization
(0CRM_IS01). Select the attribute CRM Optimization Marketing Element
(0CRM_MKTSLM), which contains the predecessor campaign element
for the optimization element. You can then drill down using the campaign element that the business partners were assigned to (0CRM_MKTELM) or using the business partner (0BPARTNER).
The DataStore object contains the predicted key figures. You can use appropriate formulas to calculate all other key figures (see Connection Between Parameters and Prediction Results).
Response Prediction Modeling
In SAP NetWeaver BI you define the response prediction models that you specify for the parameters. For more information, see the SAP Library under 
SAP NetWeaver
BI Content
Customer Relationship Management
CRM Analytics
Marketing Analyses
Response Prediction Modeling
.
You can choose from the released response prediction models in SAP CRM.
If you use response prediction models, you have to update them regularly because they are time-specific. For example, in recency, frequency, monetary value (RFM) analysis you need to perform RFM segmentation regularly on the customer master to ensure that the predictions remain as precise as possible in the planning phase as well as in the execution phase. Predictions performed using outdated data do not provide meaningful results.
You should also regularly check the effectiveness of the models.
Update the models by adding new insights.
In the case of RFM models, you need to compare the models regularly against the response rates that were actually measured.
Recalculate the RFM response rate models with current data.