Response tracking for controls
Control group responses are tracked simultaneously with offer responses, using the Response process.
Control cell responses are handled in the same way as inferred responses, except that any response codes are first discarded. For any responses from control cell members, any response tracking codes are ignored and any attributes of interest (for example, relevant products) are checked for matches against control treatment instances.
Campaign
uses an internal, globally-unique treatment code that is generated for all control treatments; however, control treatment codes are not given out, as control treatments are always no-contact, hold-out controls.
It is possible for the same event to credit both target treatment instances and control treatment instances. For example, if a particular customer is targeted with an offer for 10% of any purchase in the women's department, and that customer is also a member of a hold-out control group monitoring for any purchase from the store, if that customer makes a purchase using the coupon, that event would be associated with both the target treatment instance (using the coupon's treatment code) and the control treatment instance. Control treatment instances are also marked within the valid date range or after the expiration date, in the the same manner as target treatment instances - this provides a valid control comparison for late activity in the target cell.
Best or fractional attribution is not used for control cell responses - multiple attribution is always used. In other words, if a respondent is in a control cell for an offer and her action qualifies as an inferred response for multiple control treatments, all of these matching control treatments are credited for the response.
Copyright IBM Corporation 2015. All Rights Reserved.