I feel like i know the answer to this but I just want to confirm.
My client asked if we can AB test a segmentation strategy i had proposed on our flow campaigns but I’m pretty sure that’s not possible since most flows are triggered by an action.
Additionally, I started wondering if you can even AB test segments for like campaigns? Again I don’t think you can because essentially people can fall within the same segment so they’d get an email twice. But I’d like to confirm.
Because then I started thinking, how would I know which segment worked better for our campaigns, by just looking at the numbers?
For context my client is sending email to his whole newsletter list so I suggested we created a new segment of engaged 120.
Solved
newbie here-- can you AB test segments?
+2Best answer by DavidV
Hi
Welcome to the Klaviyo community and great question!
You can’t directly A/B test a segment itself, but you can A/B test inside of a flow. So if you had one specific condition (let’s say you wanted to test engaged 120 vs. engaged 60) you could add a profile property and create a flow filter that manually routes 60-120 into a different pathway to 0-59. This will give you the specific results based on the actual property so you can prove a hypothesis.
If you did want to use 2 segments you can always avoid the cross-over in receiving the email twice by adding a profile filter on the trigger, example:
The other functionality to call out here is that you can create a conditional split inside of a flow. So for example if you wanted to take the entire segment but you wanted to experiment with wait steps, you could build something like this:
Hope that gives you some inspiration of different ways to approach this!
Thanks
David
+32Hi
Welcome to the Klaviyo community and great question!
You can’t directly A/B test a segment itself, but you can A/B test inside of a flow. So if you had one specific condition (let’s say you wanted to test engaged 120 vs. engaged 60) you could add a profile property and create a flow filter that manually routes 60-120 into a different pathway to 0-59. This will give you the specific results based on the actual property so you can prove a hypothesis.
If you did want to use 2 segments you can always avoid the cross-over in receiving the email twice by adding a profile filter on the trigger, example:
The other functionality to call out here is that you can create a conditional split inside of a flow. So for example if you wanted to take the entire segment but you wanted to experiment with wait steps, you could build something like this:
Hope that gives you some inspiration of different ways to approach this!
Thanks
David
+2Thank you so much!
Would you say that AB testing segments is necessary for a flow? If the flow is triggered by an action wouldn’t you want it to target the person that took the action regardless of engagement levels?
Thank you I appreciate.
+32No worries
One of the triggers is “Added to segment” which you can use (so you can do advanced segmentation or have multiple actions included ie. placed an order at least 3 times AND bought product X AND spent over X etc.). The only limitation of the added to segment trigger is that a customer can only receive the flow once. See more info here.
So in short, you can have multiple flows that are triggered on segments (to see how they perform against each other), or have a broader segment and test inside the flow how subgroups of these recipients perform.
Hope that helps!
Thanks
David
Log in to the Community
Use your Klaviyo credentials
Log in with Klaviyo
Use your Klaviyo credentials
Log in with KlaviyoEnter your E-mail address. We'll send you an e-mail with instructions to reset your password.