We would like to use A/B testing in Journey Builder. I saw there was no such option out-of-the-box, but is there a workaround? When I say "A/B Testing", I don't mean just having a random split to monitor which of the 2 email versions has a better-performing e.g. subject line. Rather, we'd like a standard A/B functionality from Email Studio, whereby Marketing Cloud evaluates how each version of the e.g. subject line performed on the 20% list, and sends the "winning" version to the remainder of the list (80%) automatically.
[SalesForce] A/B testing in Journey Builder
Related Solutions
This is expected behavior. When you fire an Event (using a Fire Event Activity in an Automation or defining a schedule for the Event), a 'High Water Mark' is added to last record in the Data Extension. So, the next time the Event is fired, only new records that have been added to the Data Extension will enter the Event. Refer to this help document (specifically the 'Contacts fell below the High Water Mark' section) for further information.
However, this might be OK for your intended use. When you fire an Event, the Contacts in your Event Source Data Extension will be evaluated by the Event and providing they meet the Contact Filter Criteria (defined in your Event), then they will enter the Event.
I'm assuming that you published the new Interaction after firing the Event, in which case the High Water Mark was set on the Data Extension and the Contacts will not enter it. However, if both Interactions were published, then the Contact would enter both Interactions when the event is fired.
However, if you want all Contacts to be re-evaluated each time the Event is fired, then you will need to create a Query Activity (and include it in an Automation) to target the Event Source Data Extension and overwrite the existing records, which essentially resets the Water Mark.
Regarding the 'view event results' behavior, I'm also seeing this in the current release. You're correct in that it should indicate the percentage of rejected Contacts (that did not meet the entry criteria) and the percentage of Contacts that would have been accepted into the Interaction, but it appears that this not working at the moment.
I setup a webinar with @Keef to review the issue that he was experiencing. I was able to assist him in resolving this issue and his journey is now functioning correctly.
For the benefit of others, here's what to try if you experience the error message when publishing an Interaction:
Event.xxxxxx.xxxxxx cannot be used in the Interaction Defaults email list. Event key does not correspond to the interactions entry event key.
In @Keef's case there were a couple of root causes:
Population Configuration
The Event Source Data Extension had been added to an Attribute Group and there was no Population created. We deleted the Event Source DE from the Attribute Group (it doesn't need to be there) and we created a Population (formerly known as a 'root relationship' in earlier versions), linking a Data Extension containing all Contacts to the Contact Key of the Contact Record.
Channel Address Order
Ensure that one of the following has been configured:
- The Email Address from your Event Source Data Extension is added as a Channel Address Order in Contact Builder (from the Contacts Configuration page) or,
- In the Settings page for the Journey, select
EMAIL ADDRESS FIELDS FROM ENTRY SOURCEas the Primary Email address.
Best Answer
Yes, the most simple way is to use a Random Split.
After you randomly split the contacts, you can define the Email Activity and use the same email with different subject lines, or add a Wait activity to send it at different times.
The downside of this is that you will have to figure out a way to create reporting, which won't be as easily displayed as the one in the A/B Test in Email Studio. As a workaround, you could include Engagement Splits in your journey, just before the Exit, to see how many subscribers engaged with each version of your email.