Create a widget based experiment

This type of experiment allows you to add an experiment slot to any given layout on the Design page and configure it to run experiments at a region/widget level.

Experiments created using an experiments slot have Widget as the Targeting Method associated with the experiment, and may be the most suitable experiment type if you are running experiments for specific widget/multiple widgets on the specified layout. It is advisable to use only one experiment slot per layout.

Note: If an experiment status is ‘In Progress’, the system prevents the user from adding or removing variants, or changing the variant name. Avoid making changes to layouts or widget instances being used by an experiment as this may impact the results.

To create a widget based experiment using slots:

1. Open the Design page, and click the Layout tab.
2. Navigate to any layout by using the filter options.
3. Select your preferred layout and click the Grid View icon.
4. Locate the relevant section where you would like to create an experiment and highlight an existing region.
5. Either:

   Select an existing region and ensure all widgets are removed so that just the region itself remains.

   OR

   Add a new row (which contains regions) using the arrows on the row menu bar.

6. Open the Components menu, select Experiment Slot and drag to the empty region. You will see a default experiment slot containing a control (or default) variant and an unnamed variant.
7. Click the Create Experiment flask icon to display the configuration options with the Details tab open as default.
8. Enter your experiment name and description details in the Details tab. You can click on the View Experiment button in the Details tab to open the selected experiment in Oracle Commerce Experiments directly.
9. Select an audience and set your traffic allocation for the Control and Variant targets in the Target tab.
10. Select from a set of default goals in the Goals tab, or create custom goals by selecting Add Goal and providing the required information, such as, Goal Name, Goal Type, Commerce Event Goals, and URL. Note: The URL is required only when the goal type is Page View.
11. Start the experiment immediately by selecting the Start Now option from the Start drop down menu. You also have an option to Start on Schedule which, when selected, prompts you to enter start and end dates.
12. Once the experiment is created you can see the dashboard with the Results tab displayed. You have options to refresh, export, stop, and schedule the experiment. You can also click the View Detailed Results link to open the full details of the experiment in Oracle Commerce Experiments.

    Note: You can click Delete Experiment in all configuration tabs to delete the selected experiment.

13. You can now perform one of several options within the slot if required, as described below:
    To customize the experiment slot, click the:

    • Add variant ‘+’ button on the far right to add up to a maximum 5 variants, these can be easily re-ordered via drag and drop.
    • Remove variant ‘x’ button to the right of each variant name to remove it. You must have at least one variant to create an experiment.

      Note: You should not remove a variant if an experiment is scheduled, or in progress.

    • Variant name to open the Variant Label box where you can rename each one as required.
    To configure your experiment slot settings:

    • Click the Settings icon to display the configuration options.
    • Choose your slot name and description details.
    • Click Save to confirm your settings.
14. Save your options.
15. Drag and drop widgets from the Components menu to the control variant or renamed variant(s) as required.
16. Preview your changes before publishing.

An example of creating a widget based experiment could be to test how using two different designs on a mobile view can impact site traffic. When creating the experiment slot you can add a banner and a number of collection carousels for the control group, and add a banner and a single collection carousel for a the variant group. You could choose to target all visitors, or a specific audience. If you choose a specific audience, the whole experiment will be targeted to it. You can choose to redirect traffic equally or differently across the different groups. By default, traffic will be split equally between the different groups. Finally, you need to configure your goals and run the experiment. See Configure the target and Configure the goals for more details.

For more information on creating this type of experiment, you can filter for Experiments topics on Cloud Customer Connect where there are several How-To examples explaining step-by-step configurations, such as, Use Case #8 - Mobile Experiments.