Skip to main content

When configuring an engagement, we can specify user sample groups that should allow us to do A/B style testing. Unfortunately, that’s only half the puzzle, because we don’t have the ability to analyze the data filtered for those user groups. For example, we can tell an engagement to only show up to users in groups 1-50, but then we can’t filter for groups 1-50 when evaluating the results. Therefore, we can’t really do A/B testing or evaluate engagement effectiveness against control groups. Or at the very least, it is unnecessarily difficult.

This is CRUCIAL for evaluating engagement effectiveness, so please make the User Sample Groups available across PX for analysis and filtering 😃.


P.S. Yes I have seen this post, but I don’t think the approach it advocates is effective. Growth over a time period is not a good compare, because it will be quite different when you first launch the engagement than it will be a few weeks later when there's a slower (but steadier) trickle of users seeing the engagement. You need to compare:

  • How many users met the engagement's audience criteria but did not see the engagement? Of THOSE, how many users successfully scompleted the target action]? Or, what % ?completed the target action]?
  • How many users met the engagement's audience criteria and DID see the engagement? Of THOSE, how many users successfully scompleted the target action]? Or, what % ?completed the target action]?

Thanks for posting to PX Community @danielh !

 

You are correct, the Group #’s are not available to use as filters or visible anywhere in the UX.  This request has been opened before by customers like you, so hopefully many more users will upvote your great idea and it will be considered/added to the UX soon so it can be used as filters in PX Analytics.

 

You may have already seen this previous post too, but is is related for sure.  Some important points about the Group numbers are listed below:

  1. To be shown the Engagement, the user must meet the Engagement Audience Logic and Qualification scope AND have a resulting number within the configured sample Group number range. (e.g. 1-50)
  2. This Group # will ALWAYS be the same for a specific User or Account, it is immutable, and it does NOT change between different Engagements or anytime during the PX User/Account lifecycle.
  3. As more users qualify to view the Engagement, more users will also match this Group number ^ criteria and see the engagement.
  4. When multiple engagements use the same Group number range settings, the possible set of users that could see the Engagements based only on this Group number will be EXACTLY the same.  

To calculate the Group number, PX uses the standard Java hashCode() object method as described here. For Users, the internal "aptrinsicId" is used for this calculation, which is different from the "identifyId" that customers pass to PX.  For Accounts, the passed "accountId" value is used for this calculation. 

 

If you or anyone else is interested, below is sample Java code that generates these Group numbers and the sample Group number results for both Users and Accounts.

 

 

Using the sample code above, it is possible to generate these Group numbers for all Users/Accounts and then load them to a PX Account/User custom attribute field for use in PX Analytics filters using our PX REST API, PX Data Loader, or S3 Account and User Import features.

 

I hope I have helped a little, even if by just upvoting your idea. 

 

Happy PX-ing!

Thanks @link_black, I appreciate the info. That said, since we will constantly be getting new users added, it would have to be a recurring process for us to add the group numbers ourselves. Sounds like Gainsight already has a clear idea of how to do this, so hopefully it won’t be too much for your team to handle this for us 😉.

Reply


Terms & ConditionsCookie settings