ABX tests are one of the most useful features when choosing which strategy will work best with your customers. You can increase conversions by testing two use cases for displaying products in a web banner and identifying which is optimal.
In this article you’ll learn how to optimize product recommenders with an ABX test.
Objective: Retention.
Difficulty: Intermediate.
When to use it: When you want to compare two or more recommendation strategies.
This strategy is part of our pool of templates. To implement it in your account, access your Connectif account and go to “Workflows Create new workflow”. In the Create from template section, enter the name of the workflow in the search engine (ABX Test to optimize product recommender) then click Select andin con Create.
Video explanation
How it works
To test the product inline content, we will create an ABX test in a workflow that compares two use cases: the most visited products and the best-selling products.
Workflow configuration
Before configuring your workflow, create the following elements:
— Two pieces of inline content, to show the products of each branch of the test on the website.
STEP 1. "Start" node
(This node will set the target audience for the action).
1. Click (Edit node settings) for the “Start” node.
2. Under Select limitation, check "All my list".
3. Under Select data source, check “All existing and new”.
4. Click Update.
STEP 2. “On page visit” node
(This node will detect contacts who visit any page).
5. Click (Edit node settings) in the “On page visit” node.
6. Under Limitations, uncheck the "Deactivate after triggering" box.
7. Click Update.
If you want to learn more about node limitations, click here.
STEP 3. “Start A/B/X test” node
(This node will start the ABX test).
8. Click (Edit node settings) for the “Start A/B/X” node.
9. In Sampling stage, set the "Maximum limit of contacts to be sampled". In this case, "1000”.
10. Click Update.
STEP 4. First “Fetch products” node
(This node will fetch the best selling products of the last month).
11. Click (Edit node settings) in the “Fetch products” node.
12. To select use cases, drag “Best sellers” to the right block and click Next .
13. In Main use case settings, select "12" for Number of products to fetch.
14. Under Select the time period of the visits, choose "Best sellers in the last month" and check the "Shuffle the fetched products” box. Click Next .
15. In Summary, locate the section Number of products to fetch and indicate “12”. Next, activate "Exclude the products exported by the nodes before this one".
16. Click Complete .
If you want to learn more about the “Fetch products” node, click here.
STEP 5. First “Send web content: inline” node
(This node will send inline content with the best-selling products of the last month).
17. Click (Edit node settings) in the “Send web content” node.
18. Select the corresponding inline content and click Next .
19. In Configuration, add the appropriate selector to display the content.
20. In the Variables step, link any variables there may be.
21. Click Complete .
To learn more about inline web content, click here.
STEP 6. “On link click” node
(This node will detect the contact's click on the inline content with the best-selling products of the last month).
22. Click (Edit node settings) in the “On link click” node.
23. In Select the type of filter, choose "On clicking on a specific link in content" and select the corresponding Inline content to which it should react. Below, determine the link of the content you want it to react to.
24. Click Update.
STEP 7. First “Evaluate A/B/X branch” node
(This node will award a point to branch A if the contact's click on the indicated link is detected).
25. Click (Edit node settings) in the “Evaluate A/B/X” node.
26. Select "Literal value" and type "1".
27. Click Update.
STEP 8. Second “Fetch products” node
(This node will fetch the most visited products of the last month).
28. Click (Edit node settings) in the “Fetch products” node.
29. To select use cases, drag “Most visited products” to the right block and click Next .
30. In Main use case settings, select "10" in the Number of products to fetch.
31. Under Select the time period of the visits, choose "Most visited of the last month" and check the "Shuffle the fetched products” box. Click Next .
32. In Summary, locate the section Number of products to fetch and indicate “10”.
33. Click Complete .
STEP 9. Second “Send web content: inline” node
(This node will send inline content with the most visited products of the last month).
34. Click (Edit node settings) in the “Send web content” node.
35. Select the corresponding inline content and click Next .
36. In Configuration, add the appropriate selector to display the content.
37. In the Variables step, link any variables there may be.
38. Click Complete .
STEP 10. Second “On link click” node
(This node will detect the contact's click on the inline content with the most visited products of the last month).
39. Click (Edit node settings) in the “On link click” node.
40. In Select the type of filter, choose "On clicking on a specific link in content" and select the corresponding Inline content to which it should react. Below, determine the link of the content you want it to react to.
41. Click Update.
STEP 11. Second “Evaluate A/B/X branch” node
(This node will award a point to branch B if the contact's click on the indicated link is detected).
42. Click (Edit node settings) in the “Evaluate A/B/X” node.
43. Select "Literal value" and type "1".
44. Click Update.
Success!
Your workflow to establish an ABX test to optimize the product recommender is now ready.
Once the test has started, a sampling phase begins and, when the workflow ends, the winning branch is automatically selected. If you want to know more about ABX tests, click here.
Do you have questions?
Don’t forget, our Connectif specialists are here to help you. To contact them, just open a Support ticket by clicking the blue “Help” button on your dashboard.
Keep learning!
To make the most of your Connectif account, we recommend reading these articles next:
- Product recommender for undecided customers, to show products related to ones they have visited.
- Offer recommender, to display a banner of products with discounts.
- Recommender on product sheet, to show relevant products recommended to the one visited in the product sheet.
- Home page recommender, to show the last products visited or recommended products on the website's home page.