In A/B/n testing and optimization, conversion attribution allows you to determine which variation in an experiment is best performing based on the primary objective. Once the primary objective has been met by your visitors, the statistical engine will or will not attribute this success to one of the variations in the test. This is based on how the attribution has been configured for the particular test.
For example, if the attribution method has been set to “impressions”, the success will be attributed to the last variation that was merely “served” (i.e. presented, displayed, etc.) to the visitor, assuming that the visitor viewed the variation and that this variation played a significant role in “converting” this visitor. In other words, this variation encouraged the visitor to meet the selected primary objective.
The term Attribution Duration, refers to the length of time we want to allow the visitor’s “conversion” (having met the defined primary objective) to be attributed to the variation that was served to (or clicked by, depending on the settings) the visitor. In Dynamic Yield, the duration can be limited to the current session only or extended to 7 days or for an unlimited time, as long as the experience is still live and the visitor can still identified.