Required data

Once you know what you want to do, you must be sure that you have all the data you need to implement your plan. Optimize works with Campaign and requires the following:

*
Response tracking and analysis, including contact and response history—If you want to manage contact fatigue by ensuring you do not send too many offers to the same person, you must keep track of what you send. If you want to monitor how effective your campaigns and optimizations are, you must track customer response. You may compare the results of contacting a target group versus not contacting members of a statistically similar control group. In addition, to evaluate the effectiveness of using Optimize, you may want to hold out a group of proposed contacts that do not undergo optimization and compare that to the results of optimized contacts. Measurement of the benefit of optimization may take multiple forms (for example, increased response rate or ROI, fewer opt-outs, greater customer satisfaction, and so forth.)
*
Defined offers—You need a list of all the offers included in your optimizations as you design your rules and constraints, as these may be applied to specific groups of offers (defined as offer lists, typically based on the offer attributes or types of offers). If you plan to manually enter scores in the centralized score matrix, you need a list of the offers for which you plan to enter score values.
*
Defined segments—You must understand all the segments you want to optimize across as you design your rules and constraints, as you can constrain the applicability or scope of rules and constraints to specific segments. If you plan to manually enter scores in the centralized score matrix, you need a list of the segments for which you plan to enter score values.
*
Defined scores—As you planned your implementation, you chose a scoring method. You must have a process for generating these scores. For example, if you are populating the scoring matrix with constants, you must choose the granularity at which you will represent the scores (that is, for which offers and segments) and determine the actual score values. If you are using computations (for example, using a derived field to calculate the profitability of a proposed credit limit increase offer based on the average carried balance of an individual), you must define the equations. If you are using predictive models, you need to collect, aggregate, pre-process, and model your data in your modeling application(s).
Copyright IBM Corporation 2012. All Rights Reserved.