Run and monitor an experiment

This article describes how to run and monitor an experiment in a third-party app and change variations as needed in Microsoft Dynamics 365 Commerce.

Before you complete the steps in this article, you'll first need to publish your experiment in Commerce.

The following diagram shows all of the steps involved in setting up and running an experiment on an e-Commerce website in Dynamics 365 Commerce. Additional steps are covered in separate articles.

Experimentation user journey - Run & Monitor.

After you publish your variations, all of the steps you need to do in Commerce to run your experiment are complete. The next step is determining which variation to show to each user when they request a page. The third-party service makes that determination, but first you have to activate the experiment within the service. Since the steps for activating an experiment vary from service to service, you'll need to follow the instructions included with your service or provider. If the experiment is not activated, users will only see the default version of the page (no variations will be displayed).

You'll need to keep the experiment running long enough to gather data for statistically valid results. Use the third-party service to monitor the experiment-related data and analytics while the experiment is running.

Adjust your variations

If for any reason you need to modify your variations, follow the steps below.

Important

If you make changes to a live experiment in Commerce or the third-party service, your results may be significantly impacted. Consider letting the experiment run its course and then creating a new experiment for major changes.

  1. In Commerce site builder, select Experiments in the left navigation pane, and then select the experiment.
  2. Select the variation you want to update from the drop-down menu.
  3. Make any needed changes, and then preview and publish the variations. For more information, see Preview and publish an experiment.
  4. Go to the third-party service to make any experiment setup-related changes.

Previous step

Preview and publish an experiment

Next step

Promote a variation and complete an experiment