Directional experiments let you test assets or groups of assets head-to-head in a controlled environment. You can select different assets, create distinct test groups, and choose a success metric relevant to your App campaigns for installs (ACi).
This article explains how to set up Directional experiments for App campaigns and includes best practices you can follow. Learn more About Directional experiments for App campaigns.
On this page
- Prepare for success: Prerequisites to a successful experiment
- How to set up a Directional experiment
- How to check Directional experiments results
- Best practices
Prepare for success: Prerequisites to a successful experiment
To help you set up a successful Directional experiment for your App campaign, consider the following:
- Bid higher or more competitively to ensure that your test group gets enough traffic.
- Use Maximize conversions bid strategies for App campaigns for faster results, though it may be more costly.
- Create a new video-only App campaigns for installs (ACi) with exactly one active ad group that contains mandatory assets and all the video assets (minimum of 2 videos) under one test group.
- Note: There should also not be any other active test in the same campaign.
How to set up a Directional experiment
- Go to Experiments within the Campaigns menu.
- At the top of the “Experiment” table, select the Custom experiment tab, then select the plus button
.
- In the “Choose an experiment type” page, select Assets in the “What do you want to test?” card.
- In the “Choose a variable to test” card, select Assets provided by you.
- In the “Campaign type” card, select App, and in the “Experiment type” card, select Directional split.
- Select Continue to proceed.
- In the “Test Campaign” card, click + Select campaign. In the dialogue box that opens, choose the campaign that you want to test.
- Note: When you choose a campaign, it will show a read-only preview of the assets that are available in your campaign's ad group.
- In the “Test groups” card, select + Add assets. In the slider panel, select the video assets that you want to add in your first test group, then select Save.
- Note: Video assets that aren’t selected for a specific test group will be used across all test groups.
- Select Add test group to create additional test groups.
- Note: The option to add test groups will be disabled when all eligible video assets have been assigned or if you've reached the maximum limit.
- In the “Success metric” card, choose a success metric from the Select metric drop-down.
- In the “Experiment name” and “Experiment dates” cards, enter an experiment name and select the date range for your experiment.
- Note: Test results can be viewed within 2-3 days. The default end date for a test is 14 days after the start date.
- Select Create experiment.
How to view Directional experiments results
To check your experiment’s progress or end results, follow these steps:
- Go to Experiments within the Campaigns menu.
- At the top of the “Experiment” table, select the Directional experiments tab.
- Select your experiment's name to view performance details. You will be presented with a summary of the top-performing test group based on your chosen success metric.
- Note: It’s possible that the top-performing test group may change as the experiment continues.
Best practices
- Directional experiments are designed to provide faster directional results. However, there are no strict recommendations for test duration or budget. We encourage you to determine these based on your specific testing methodology, Key Performance Indicator (KPI) benchmarks, and campaign goals.
- To help you get the most valuable insights, we recommend that you test only one attribute in a single test. For example, you should test a green vertical video against a red vertical video to compare color. If you test a green vertical video against a red horizontal video, the test will compare both color and orientation, making it harder to get a clear read on performance.
- For future tests, simply create a new test group within the same test campaign. This helps you reuse the campaign while minimizing performance variations that might occur when you start a new one.
