Smart Campaign Multiple Field Dependent A/B Testing

[Daniel_Tien]

I'm trying to set up an A/B email test in a Smart Campaign.

In a simple 50/50 A/B email test, it looks like I just need to create an "Add to List" step at the beginning of the campaign flow to separate the 50/50 random sample, and then designate at the "Send Email" step which email each group should receive based on their list membership.

But if I need to create that 50/50 random sampling based on multiple data field filters (ie. I'd like to create a 50/50 split of female members who live in California, and then a 50/50 split of female members who live in New York), is that possible to do in the same Smart Campaign? I'm playing around with the "Add to List" flow and it doesn't look like I can add more than 1 filter condition to each "If" step.

Or is there a better method for this type of segmentation?

[SanfordWhiteman]

You can create a Segmentation w/4 Segments (given that in this case you have so few permutations, in more complex cases segments would be unreasonable/impossible to precompute) and then base your choices on that. 

[Daniel_Tien]

I'm able to create segments:

1) Female Users in CA

2) Female Users in NY

I might just be missing something, but it doesn't look like I can create the 50/50 random sampling within the segments, or in the smart campaign once those segments are created. 

[SanfordWhiteman]

Find the segment in the Add Choice, just as you would find a field.

[Daniel_Tien]

I'm able to find the segments in "Add Choice", but I'm not able to layer on the "Random Sample is 50%" filter (I can't do IF "Segment is XYZ" AND "Random Sample is 50%", Add to XYZ List.)

[SanfordWhiteman · Accepted solution]

Sorry, disregard the above, it was going to get too complex. No more bright ideas while I'm on an interstate bus.

Instead do it with multiple Smart Campaigns: first Add to List, then split again by Member of each List.

[Daniel_Tien]

I was hoping that wouldn't be the case. 

Safe travels!

[SanfordWhiteman]

Thanks!

As you get deeper into Marketo, you'll probably find that Random Sample is too erratic for people's typical use. What you're likely to want at some point is true round-robin, or true randomness, both of which can be done using webhooks.