Multivariate Test best practices

Be aware of the locations on your page that are likely to produce significant results.

For example, a banner or a hero image is probably going to lead to more conversions than a change in the footer. Including less-influential locations in your test only increases the amount of traffic and time required to test the more prominent locations on the page.

Prepare your page variations ahead of time.

Be aware of the content differences for each offer, and create any images, text, and HTML offers that you expect to use in the MVT test.

Don’t include more combinations than necessary for the test.

Every combination tested significantly increases the amount of traffic and time required to achieve acceptable results. For example, if you have three locations with three offers each, there are 27 possible combinations (3x3x3). Three locations, where two locations contain three possible offers and one location has two offers, provide 18 options (3x3x2). The numbers increase substantially with each additional location and offer.

Name locations and offers.

You can rename each location and offer in your test to something more useful. The number of offers in each location appears in the location header. Useful names help you identify your offers when examining reports.

Take advantage of the preview features to avoid undesirable combinations of content.

Review all the experiences generated by your test before going live. Make sure that there are no combinations with contradictory claims (for example, 20% off and $19 off in the same experience) or incompatible designs such as having background and font of the same color.

Use the Traffic Estimator to make sure that your test is designed for the amount of traffic your page receives.

Make sure that the Traffic Estimator gives your test configuration the green light so you can get the results you desire.

Each element’s alternatives should be significantly different from each other.

Make frequent use of the Location Contribution report to monitor the performance of each location and offer.

In the Experience Performance report, base your decisions on the data shown using the Best 5 and Worst 5 filters.

The All filter makes it difficult to extract the desired information, and not all experiences can display in the graph. Use the All filter if you want to look at a specific experience that is not in the best or worst five.

Although Target allows you to edit a live activity, editing an activity that is in progress could reset the test. Reports might not recognize some of the changes. It is safe to make changes to HTML offers in offer library only.

Specific actions that reset experience names and reports include:

• Adding a new location
• Deleting a location
• Adding new offers or deleting offers from an existing location
• Editing rich text offers
• Editing background color offers

By following an MVT test with one or more A/B tests, you can determine the best possible content for the results you desire.

Once you have determined which locations and content are most useful for helping you meet your goals, you can run an A/B test to further refine the results. For example, when you know which locations are most important, test two specific images against each other, or compare the wording or colors of a call to action.