Forum Discussion

JudgeJules's avatar
JudgeJules
Active Member
3 months ago

Randomizing NPS messages

Randomization seems harder than it should be in Braze.  I've got a canvas scheduled to run everyday at the same time and send a custom in app message that will collect an NPS score for those that interact with it. We are targeting all users in our system (over 5 million). 

The controls are: "Users are not able to re-enter this canvas. Only 500 users will be able to enter this canvas everytime this canvas is scheduled"

We've yet to start this -- I'm concerned on the outcome.  Will this randomly select 500 people to send to daily and never send it to them again?  Key here is Random...and does it apply to a scheduled canvas that sends regularly?  Ideally we'd be able to randomize based upon those who have a session and trigger it off that.

Our requirements are such that users must be randomly selected, cannot receive it more than once and hopefully send it sometime after a session begins. I've looked at targeting random bucket numbers and using a session start as the trigger...but I think this just keeps the target the same and will only let those users in the canvas.  

Has anyone found a way to do this?  I wish there were a randomization step in canvas, that would make this super easy.

Canvas

3 Replies

Sort By
  • JudgeJules's avatar
    JudgeJules
    Active Member
    3 months ago

    Thanks for the reply inespais. It seems with many efforts there are multiple ways and this may be making a square peg go into a round hold.  Braze does many things and I'm not sure this is one of them (yet!).

    We've got a partial solution though limited in capability...though not super easy to find.

    We went with a scheduled canvas and used the canvas entry limit of X (500 for us), per scheduled period (ours is daily), which I hadn't seen before--this allows a functional random selector when Entry Controls are applied.  

     

    The canvas evaluates users each day at the same time and selects a random 500 to enter and then cannot re-enter (unless you wanted them to).  To increase impressions occurring we targeted a segment of users who've had a session in the last 4 weeks and are not "new" in the last 4 weeks. This could be changed for any specific requirements of who to target. In our case we can live with the built in bias of users already being "active" when they receive the NPS.

    Fun side outcome is we've set this up to then pass the results to a webhook sending over to our CDP where anyone in the company can view/analyze.  I love when integrated systems just work!

    Hope this help anyone trying something similar!

  • inespais's avatar
    inespais
    Practitioner II
    3 months ago

    Very cool! Thanks for sharing JudgeJules - to be honest it’s the first time I see that as well 😅 but I never had the need to limit entries like this and I guess it’s a tiny feature you’d only find when you really need it.

    The webhook part makes a lot of sense. Not sure if you’re doing it already, but you can also store the user’s response as a custom attribute and later re-target them: say you had an NPS score range between 1-5 - you could retarget those that chose 4 or 5 with a prompt to rate the app on the store or share with a friend; and those that chose 1-3 with a feedback survey or so.

    Good luck with the NPS!

  • inespais's avatar
    inespais
    Practitioner II
    3 months ago

    Hey JudgeJules I don't have an immediate solution but would it be an option to use a campaign instead of a canvas - you might have more control that way.

    My understanding is that within canvases, IAMs are delivered to the device as soon as segmentation criteria is met and remain "on hold" until a session is started - this would be worth double checking with your CSM rep, but if it's correct it might indeed make it quite hard to select users, as you can't guess which 500 will open the app on any given day. I think this behavior is slightly different for campaigns as you could do action-triggered based on session start instead of a schedule.