Optimize your Automations with Workflow Split Tests
Split tests are a powerful tool to test out your marketing strategies and to determine which strategy performs better. Drip makes split testing more powerful than ever with our Workflow Split Tests.
Workflow Split Tests go a step further than telling you which email performed better, by showing you the direct and tangible impact of your marketing strategies to your brand's bottom line. In the guide, we’ll go over how to conduct a split test to get the most out of Workflow Split Tests.
Learn how to split test subject lines and email content for Broadcast emails in our User Manual.
What’s My Goal?
Clearly define your goal and what your measurement of success will be. There are four measurements available in Split Tests: Revenue, Order Conversion, Site Visits, Email engagement.
Say your goal is to promote your semi-annual sale. You would measure the success of that goal with revenue. If your goal is to encourage people to complete the checkout process, you would measure that by order conversions. Here are a few other examples of goals for each measurement you can test out:
Here are some examples of goals for each measurement that you can test:
- Goal: Post-purchase follow-up | Measurement: Site Visits
- Goal: Announce a new product by testing email content | Measurement: Email engagement
Clearly defining a goal and how you will measure the success of that goal helps keep your test focused on measuring one thing vs the other to see which has the most desirable outcome.
What do I want to Test?
With Workflow Split Tests, we recommend limiting the number of variables you’re testing in a single split test. Limiting the number of variables allows you to clearly attribute success to a given variable. Here are some of the variables you can test with Drip’s Workflow Split Testing:
You will want to have a significant number of people in each variation, especially for email tests. Manually declaring the winner of a split test allows you to track the progress of each split test path and determine a winner when it seems that one path is indisputably performing better than the others.
Here are a couple of good resources to help you build your split tests:
Drip will not calculate the statistical significance of your split test result.
Conduct the Split Test
Once you’ve determined your goal, measurement of success and the variables you want to test, you’re ready to conduct the test.
Take a look at the Cart Abandonment split test Workflow above. Here, our goal is to get people to complete the checkout process by purchasing the product(s) they abandoned. We’re testing this by adding two variables: channels (Facebook Custom Audiences) in Path B, and frequency (second email after 9 hours) under Path C. Notice how there is one variable per path in the split test. This allows us to have a clear understanding of which strategy to use moving forward when we evaluate the split test results and declare a winning path. Keep this strategy in mind as you’re creating your Workflow Split Test.
To create a split test:
- Add an action and select the Split Test node
- Name the split test
- From the drop-down, select the number of paths to measure against one another
- Click Save
- Under each Path A/B/C/D/E, add the node(s) needed to complete the path
- Select an Action, Decision, Parallel Paths, or Delay in a nested Split Test Workflow node
- Click Start Workflow
Evaluate Split Test Results
Use the Results Modal to evaluate the metrics and declare a winner. Here, you can see the following metrics:
- People represent the total number of people in the path
- Order Conversion is the number of orders attributed to people in the path divided by the number of people in the path
- Revenue per Person is the total number of attributed orders divided by people in the path
- Site Visits is a gross count of sessions
Look at the measurement that you used to measure your goal to help you determine the winning path.
For example, in the Cart Abandonment split test, we would select the winning path based on our goal of getting people to complete the checkout process by purchasing the products they abandoned. Our measurement for that goal was order conversions, so we would select the path with the highest number of order versions as the winning path.
To manually declare a winner:
- Click the Split Test node
- Click the Declare Winner button for the winning path
- The Workflow will now display only the winning path
Now, everyone who hits that point in the Workflow will go through the winning path.
Moving forward, you can confidently send everyone down this path in the Workflow and leverage this knowledge in other split tests. Say frequency leads to the highest number of order conversions, but not to the highest number of site visits. You can leave frequency out and test with other variables when you’re measuring site visits.