Hi PX-ers!

We continue to see much interest in leveraging the powerful Engagement Sampling options in PX, especially with our customers that engage in-app with large amounts of application users.  

Not only does Engagement sampling provide a way to limit how many users actually see the Engagements/Surveys, but can be used to spread out Engagement views and Survey responses more evenly over time. 

I recently posted another tip for implementing rolling NPS Surveys that you may want to review as well.

To help you better understand how the PX Engagement sampling works, I wanted to provide some additional details to help you to decide the sampling option that will work best for your use case.

The Basics

To begin with, the Engagement Sampling options are available in the Schedule -> Qualification Scope settings of an Engagement configuration and there are two options: Engage % of the targeted audience OR Engage a group Users/Accounts based on their assigned Group #.

The Engagement sampling setting chosen above will determine how many users will be able to view the active Engagement, but do remember that to view any Engagement a user must match both the Audience Logic and configured Qualification scope settings, as well as adhere to any Engagement Throttling settings that may be in play too. So, they need to be “fully qualified” to see the Engagement before any of your Engagement sampling criteria is even considered.

The Details

Let me explain these two Engagement Sampling options in greater detail…

Option 1

The Engage % of the targeted audience default setting is 100%, so any actively tracked users that have matched the configured Engagement Audience Logic, Schedule, and other Qualification scope will view this Engagement.  If you change the default % to something less, then you have effectively reduced the number of users that can potentially view the Engagement.

When a different sample % is used, then PX uses a math formula to calculate a number between 1-100 for “fully qualified users”, which is compared to the configured sample %.  To be shown the Engagement, the user must have a resulting number less than or equal to the configured sample %.

Below is the simplified math formula:

some_value -> amazing math formula = a number between 1 and 100.

It is also important to note that Engagement Sampling % works differently depending on the Engagement Qualification Scope settings too.  

Using “Engage 30% of the targeted audience” as an example…

only once:  ~30% of all tracked users that match the audience logic will see the Engagement.  These users either qualify or they do not based on a math formula using EngagementID+UserID, so it is always a fixed # of users that qualify per Engagement. New users could also qualify, but any existing user that did not qualify before will not qualify later. 

• If you increase the % over time then more users will qualify to see the Engagement.  
• And, since the sampling math formula includes EngagementID, ~30% of tracked users for one Engagement will lead to a different ~30% of tracked users in another Engagement that will see the Engagement.

once per visit: ~30% of the sessions for tracked users that match the audience logic will see the Engagement.  This means that any user can qualify to see the Engagement using a math formula using SessionId.  So, this controls how many users actually see the Engagement on a given day/week/month, but it could be any user.  

• This is a great way to "scatter" Engagement views and Survey views/responses over time.
• It is also possible that some users may never have sessions that are inclusive of the sample % and will never see the Engagement.

intervals: ~30% of the sessions for tracked users that match the audience logic will see the Engagement.  This means that any user can qualify to see the Engagement using a math formula using SessionId.  So, this controls how many users actually see the Engagement on a given day/week/month, but it could be any user.  

• This is another great way to "scatter" Engagement views and Survey views/responses over time.
• It is also possible that some users may never have sessions that are inclusive of the sample % and will therefore never see the Engagement.
• And, since "intervals" replays the Engagements for users, this also ensures that ~30% of the subsequent sessions for users that match the audience logic AND already saw the previous view/occurrence will view the replays of the Engagement. However, because of this, there is no guarantee that a user that saw the first view/occurrence will always see all the subsequent views/occurrences. 

every time (paywall): ~30% of the tracked users that match the audience logic will see the Engagement.  However, this is determined by PX choosing a random value between 1 to 100 for any qualified user at runtime, which must be inclusive of the sample % for the Engagement to be shown. This means that any user can qualify to see the Engagement using that randomly selected value.  So, this controls how many users actually see the Engagement on a given day/week/month, but it could be any user.

• This is also another great way to "scatter" Engagement views and Survey views/responses over time.
• It is also possible, but unlikely for frequently tracked users, that some users may never get a random value assigned at runtime that is inclusive of the sample % and therefore will never see the Engagement.

Option 2

The “Engage a group of Users/Accounts” by their assigned group numbers between 1-100 is a great way to reduce how many users/accounts will see the Engagement and is invaluable in A/B testing as I previously posted here.  

This Group # will ALWAYS be the same for a specific User and Account, they are immutable and will NOT change. Therefore, when the same Group number criteria is used in different Engagements, the set of Users/Accounts will always be the same.  And, when different Engagements use different Group number criteria, the set of Users/Accounts will be different.

While you cannot currently see these “Group” numbers in the PX application for Users or Accounts, they really do exist and are calculated using the simplified math formula below:

User Group ID: (User Id) -> amazing math formula = a number between 1 and 100.

User Group ID: (Account Id) -> amazing math formula = a number between 1 and 100.

To be shown the Engagement, the user must be “fully qualified” to see the Engagement AND have a resulting number within the configured Group number range. (e.g. 1-50)

The (short) Summary

You can choose to use either of these Sampling options to limit the amount as well as the frequency of Users/Accounts that will view your amazing Engagements/Surveys. So, think about your Engagement sampling use case and pick the one that will work best for you.

Happy PX-ing!!!