Subject lines: Everything you know may be wrong. Part 2–setting up the test

The week before last we asked whether simplification of our email newsletter’s subject line might result in a greater open rate. My theory was that newsletters, as they are pre-sold, don’t need to promote themselves as much. To test this I ran our email list through an excel file, added a custom column for title number, then created a dynamic subject line based on each. Ends up this process was far too complicated, and instead of working, it resulted in a 15% lower open rate (for some reason) and totally jumbled results.

So instead, David Eckman gave me the following advice on how to set up an A/B test for your newsletter. This is intended for ExactTarget, but the basic idea should work in whatever mailing system you’re using.

What we’re going to do is make two basic groups, randomly chosen from your subscribers.

1. Under users, click “my groups”, then “create group” to create a new group.
   
2. From here, you are going to create a random group from your subscriber list. Pick “pick a random group from your subscriber list”
   
3. Click add, then in the next window give your group a name and size. After you’ve done that, click add again and create a second group, with the same size. When you’ve made both, hit “create”.
   
   

That’s it. Your group should be made, and you should be ready to go.

Possible problem: When it selects a random sample it doesn’t assure that each group will have the same number of active/bounce/unsubscribed subscribers. As fate would have it, ours came out pretty close to even between the two (15 fewer actives in one, with slightly more unsubscribed). This shouldn’t be a problem, but if you have large discrepancies there are a few work arounds. I’ll get to those in my next post…