Solved

Can I make an A/B test for TIME DELAY in a flow?


Badge

How do I make a simple A/B test for the TIME DELAY in a FLOW? I can’t seem to see that it is even possible. For example, the Browse Abandonment flow, I would like to test a faster delay to send, say 15 minutes, and a slow delay, say 2 hours?

Is it even possible?

 

Thanks in advance -

icon

Best answer by Dov 22 July 2021, 15:39

Hello @Lavender_Brown,

Thank you for sharing this with the Community.

This is a great question! While there isn’t an inherent “A/B Test” option for a time delay, we can use a conditional split to effectively A/B test two discrete time-delays.

Use a conditional split containing a 50% random sample and insert the two different time-delays below this split. Lastly, clone the flow email in question and insert it under the other time delay. The final product will look something like the following image:

For more information, I recommend checking out our article on this topic here.

Thanks and have a great day.

View original

3 replies

Userlevel 5
Badge +4

Hello @Lavender_Brown,

Thank you for sharing this with the Community.

This is a great question! While there isn’t an inherent “A/B Test” option for a time delay, we can use a conditional split to effectively A/B test two discrete time-delays.

Use a conditional split containing a 50% random sample and insert the two different time-delays below this split. Lastly, clone the flow email in question and insert it under the other time delay. The final product will look something like the following image:

For more information, I recommend checking out our article on this topic here.

Thanks and have a great day.

Badge

I am new to Klaviyo and blown away by the features and support! I appreciate the answer supplied but would suggest this as a feature in the future. The simplicity of designing a flow and then having someone else ask what is the best time delay (when I should have thought to ask)...simply turning that branch into an A/B - with your ability to let the A/B decide the better outcome, set a deadline to end the test and have the best result implemented. Set it and forget it :-)  Just a suggestion! 

Thanks for the answer so quickly -

Userlevel 5
Badge +4

Hi @Lavender_Brown,

Thank you for your reply.

Glad to hear that the answer was helpful. Of course, I certainly understand how a defined feature would be helpful for collecting data more effectively. I have submitted a feature proposal on your behalf. If there is enough widespread interest, the team will start developing the feature but for now this workaround is the best way to effectively run an A/B test for splits :)

 

Reply