There are a few articles in the community about the fact that you can't use Champion/Challenger testing on batch email campaigns. We came across this when using programs in our engagement engine - misread the article about using Champion/Challenger in engagement engine (ie can on a simple email, but not a program). We use programs in Engagement Programs for multi-touch attribution analysis.
We simply made the batch that was being used in the stream to have "Request Campaign" in the flow, with a trigger campaign in the program being used listening out for the request. We then set up that trigger campaign to send the email required for the stream - since when using programs Marketo is looking for program membership we then just have some camapigns to update status based on sent/del/open/clicked, so they won't get the same email on the next cast.
We're now getting A/B testing on any of our nurture programs we want.
It's worked for us, but before you try it out I recommend you test and use a sandbox if you have one.
Mark Farnell interesting. In the past, I have some challenges using smart campaigns within a program embedded in an Engagement Program. Great to hear that this method works!
Optimizely's sort of nerdy with Statistical Significance, so we create two emails for our AB test, then use "If Random Sample is x, send Email A, default choice Send Email B". Then we throw our results into our AB testing excel doc to calculate significance.
Definitely more arduous than the Email Program, but another method nonetheless!