The Dynamic 365 2019 Wave 2 Update is now GA. At Barhead we’ve been reviewing the new features in D365 For Marketing – here’s a quick look at one of them – the new A/B Testing functionality.
A/B testing addresses the business challenge of optimising your results by incrementally testing variations of the content. It’s a key tool in the marketers toolkit.
In the current version of D365fM it’s perfectly possible to construct an A/B test. But only manually: create the A & B versions of the email, add a splitter tile to the journey and add each email variant to a track. Go live, then manually evaluate the results to determine the winner and reconfigure the journey to use that version. Your journey would look something like this:
A/B Testing in D365fM – Pre 2019 Wave2
With Wave 2 Microsoft are introducing a much better A/B testing approach, with in-email A/B testing and automated winner detection and allocation. Build one version of your email, add your A/B testing variants, and when the winner emerges, all remaining sends automatically get this variant.
Here’s a quick scamper through how this works:
Create a new email and one of the first things you’ll see is an A/B test button on the email header ribbon:
Expand this button, click +New and select from Subject line or Sender (‘From name’) tests.
For this first look I’m just going to try a Subject line test – with and without personalisation
Adding the A& B variants is very straightforward. Here’s variant A with a personalised Subject line
And variant B with no personalisation
You can add more test variants – and then select which one to test in the Journey Designer.
Test and go-live with this email, then head into the Journey Designer. Add your email that includes the A/B test and you’ll notice the email tile has small blue A/B icons in the top corner.
Activate the A/B testing slider, and select your test parameters. Key parameters include
- Which test to execute (e.g. Subject personalisation test);
- The A/B test sample sizes (in this case 10% + 10%);
- The winning metric (e.g. email clicked);
- Test duration (in this case 6 hours)
- What to do if initial test is inconclusive
If you want to revisit the email, click on the Edit button.
When you’ve configured and tested the journey, go live and the email sends will start.
For this test I used a seed list of 11s. Here’s a sample of version A, showing the personalisation:
And here’s Version B:
Once the emails start flowing, you can quickly view the KPIs – including the ‘winning’ variant. As you’ll see, of the 11 contacts in the seed list only 2 have sent (the 2 examples above, plus our hard bounce account in the seed list); the rest are on hold waiting for a winner (or the ‘test inconclusive’ time period to pass).
For this test, we opened & clicked only one of the emails, to confirm that the auto-selection of the winner would work. And work it did – just as expected. The remaining 7 emails received the winning version, six hours after the test went live.
For another review, check out this blog post by Jesper Osgaard, one of Microsoft leading SMEs on Dynamics 365 For Marketing.
In summary: admittedly a contrived test – but overall I’m impressed with both the user experience and also the outcomes. There are a few areas that I’m hoping will be addressed before GA in October – for example it’s not possible to see which version of the test a contact received – but overall, pretty impressive.