Function documentationUsing RFM Analysis in Segmentation

 

Note Note

It is only possible to use the RFM analysis for classic segmentation.

End of the note.

You can apply the results of an RFM analysis to optimize a campaign. In this way, you can restrict the target group addressed by a campaign to just those business partners with a relatively high probability of responding to the campaign. RFM analysis can be applied in this way to refine your target group.

You can restrict the target group in segmentation by using the response probability of a business partner as an attribute. The response probability is determined in SAP NetWeaver Business Warehouse (SAP NetWeaver BW). It is dependent on the type of offer that is to be promoted in a campaign and on the channel with which the campaign is to be performed. For this reason, you can create different response rate models for separate sets of requirements in SAP NetWeaver BW and then use these models as the basis for different attributes in segmentation.

Prerequisites

Settings in SAP NetWeaver BW
Customizing Settings in SAP CRM

To be able to use the results of the RFM analysis in segmentation, you need to create a data source with particular settings in Customizing and then assign this data source to an attribute list. To do so, proceed as follows:

  1. In Customizing for Customer Relationship Management (CRM), choose Start of the navigation path Marketing Next navigation step Segmentation Next navigation step Maintain Data Sources and Attribute Lists End of the navigation path.

  2. Choose Create Data Source.

  3. Make the following settings and save your data source:

    • Origin Type: InfoSet

    • InfoSet: RFM Analysis (CRM_MKTTG_RFM)

    • Description: Enter an appropriate description, such as RFM Analysis

    • Business Partner: CRM Marketing: GUID for a Member of a Target Group (CRMT_MKTTG_RFM_RESPONSE-TG_MEMBER_GUID)

  4. On the Find tab page, select an attribute list to which you would like to add one or more RFM attributes. For example, you can add an attribute that provides the probability of a response to a mailing campaign and another one for the probability of a response to a telesales campaign.

  5. To include an attribute in the attribute list, choose Assign Data Source.

  6. Select the data source that you created in steps 2 and 3.

    The data source is now included in the attribute list.

  7. Expand the node that has just been added and select the attribute CRM Marketing: Response Probability of a Customer.

  8. In the context menu for this attribute (accessible by using the secondary mouse button), choose Properties.

  9. On the Attribute Properties tab page, make the following settings:

    • Change the description according to how the attribute is used, such as “RFM for Mailing” or “RFM for Telesales”.

    • Select the attribute type RFM ANALYSIS. The attribute type determines the graphical display in segmentation.

  10. In the context menu (secondary mouse button) choose Edit Filter Conditions.

    1. Enter the following variables:

      • CRM Marketing: RFC Connection BW (CRMT_MKTTG_RFM_RESPONSE-RFC_DEST)

      • CRM Marketing: RFM Segmentation Model (CRMT_MKTTG_RFM_RESPONSE-SEGMENT_VERSION)

      • CRM Marketing: RFM Response Rate Model (CRMT_MKTTG_RFM_RESPONSE-FORECAST_VERSION)

      These variables are used by SAP CRM to apply the settings for RFM analysis that you have made in SAP NetWeaver BW (see the above section Settings in SAP NetWeaver BW).

    2. For each variable, choose I — Range limit included in the Range Limit column and EQ — Equal (= from) in the Option column.

    3. Assign the appropriate value to the respective variable:

      • Enter the RFC connection to SAP NetWeaver BW in which you have made your settings for RFM analysis.

      • Enter the RFM segmentation model that you created in SAP NetWeaver BW.

      • Enter the RFM response rate model that you created in SAP NetWeaver BW.

  11. In the context menu for the attribute you renamed in step 9, choose the Create Filter option.

  12. Set up filters for the response probability based on your experience and using the categories low, medium, or high. These filters are used to divide up the business partners provisionally. You can call up this rough division in segmentation under Distribution. Later on, in segmentation, you can make user-dependent changes to these filters and can create additional filters.

    Note Note

    Set up another filter and call it something to the effect of “No prediction possible” and limit it to all values under zero. This is necessary because the system allocates a response probability of -1% to all business partners for whom there is no RFM segmentation available.

    You could create filters as follows:

    • No prediction possible < 0 %

    • Low 0 through 1 %

    • Medium 1 through 2.5 %

    • High > 2.5 %

    End of the note.

  13. If you want to include additional RFM attributes in your attribute list, start again at step 5. You usually select a different RFM response rate model for a different attribute (see step 9).

Features

In segmentation, when you select the attribute list that has been set up using the above steps, the response probability attributes that you added are available for restricting the target group. When you select an attribute, a detail screen appears, displaying the distribution of the target group according to the filter values. Using a dropdown list box, you can call up prediction and simulation screens, which enable you to find out the potential effect that different target group restrictions have on respective key figures. Beside the graphic on each page, there is an extensive details list showing all the extrapolated values.

You can use the following pages:

  • Distribution

    This page shows how many business partners appear in the different filters.

  • Profit Simulation

    This page allows you to estimate the profit made with a campaign addressing a given target group. Here you specify the anticipated costs incurred by addressing a business partner and the anticipated profit from each positive reaction. As an extra option, you can set fixed costs for carrying out the campaign.

    The system can now use the extrapolated response rates from the RFM analysis to estimate the overall profitability of a campaign addressing this target group and assuming the anticipated figures. Furthermore, the system proposes a filter that you can apply to the attribute to increase profitability and it estimates the optimum profitability with the given figures.

    The system uses a graphical display to show in detail the effect that the different filter restrictions have on the attribute. The possible filter values can also be varied according to a “greater than/equal to” condition: a piecewise reduction of the current target group is simulated by eliminating the business partners with the smallest response probability, and the corresponding estimated response rates/campaign profit are shown. You can change the costs and profit figures interactively so that the extrapolated effects are immediately visible, giving you an insight into how sensitive the target group is to these parameters.

  • Return on Investment (ROI)

    On this page, you can perform the same simulation for the target value Return On Investment. The campaign execution costs (that is, the sum of the fixed costs and the variable costs multiplied by the number of business partners addressed) represent the amount invested in the campaign. This usually produces a different optimum filter value determined by the system. The display and the interaction options are the same as on the Profit Simulation page.

  • Number of Business Partners

    On this page, the effect of different filter options on the number of business partners in the target group is displayed graphically. If you want to address a particular number of business partners, you can select one of the filters that leads approximately to the desired target group size. You can also check the effects the filters have on profitability and on the ROI.

  • Number of Responses

    On this page, the anticipated number of responses for each of the different filters applied is displayed graphically.

More Information

Example: Using RFM Analysis in Segmentation