Hi
We are looking to test two packages in our upcoming direct mail solicitation and I am wondering how other organizations track their testing. My initial thought on how to track the different test packages was to break each segment (which has its own source code) into two segments (each with their own source code). However, the problem with doing it that way, is that there is no way to easily see just how the test package performed vs how the traditional package performed - we would need to remember which source codes were test and which were not and run reports on those specific source codes. We also would not have an easy way of looking at each segment as a whole.
Does anyone have a better way of tracking this?
Thanks
Jess Levy
San Francisco Opera
Hi Jess,
This kind of testing is exactly what sources and extractions were designed to do.
You are on the right track by breaking your two segments into 4 and creating a unique source for each. It shouldn't be difficult to know the purpose of each source, you just have to name them clearly, something like Segment A Package A and Segment A Package B. As long as you promote the sources it will be easy to select the right source when you process the contributions, because each constituent will only get assigned to one segment/source and that source will be the one displayed in the source window. If you want to roll the segments back together, you can use source group; create one source group for Segment A and one for Segment B.
To report on the results, use the Revenue by Source and Appeal report. It will list all sources side by side, and give you additional options for grouping the sources by source group, making it easy to see how the different packages perform and how a segment as a whole performances.
One other tip, when you are setting up your extractions, you can use the Random Select % feature to randomly distribute constituents in a solicitation segment between the regular package extraction segment and the test package extraction segment. If you need help with Random Select, see this help topic.
-Kevin Sheehan