Run Experiments

Info: This feature in Bloomreach Content requires a standard or premium license. Contact Bloomreach for licensing details.

Experiments in Bloomreach Content let you evaluate whether changes to a page component increase visitor conversions to a specific goal. You can test both the content displayed by a component and the audience it targets. While an experiment is active, visitors are shown different variants of a page component. The system tracks conversions for each variant and adjusts variant distribution based on visitor data. When enough data has been collected, you can complete the experiment—typically when one variant demonstrates a significantly higher conversion rate than the others.

Tasks

• Manage Goals
• Create an Experiment
• Monitor an Experiment

Concepts

Experiment

An experiment evaluates whether modifying a page component—by creating one or more variants—improves conversion. You can test multiple variants within a single experiment. Each experiment is either a content experiment or an audience experiment.

Content Experiment

A content experiment measures if changing a page component’s configuration—such as its content or presentation—improves conversion compared to the default configuration. This approach is similar to an A/B test, but Bloomreach Content supports more than two variants and uses an advanced algorithm for optimization.

Audience Experiment

An audience experiment tests whether targeting a page component to a specific audience segment results in better conversion than serving it to all visitors.

Page Component

A page component is a configurable section of a web page managed through the Experience manager.

Variant

A variant is an alternative configuration of a page component. In a content experiment, variants differ by content or presentation. In an audience experiment, variants are targeted to different audience segments.

Goal

A goal is a specific page that a visitor must reach after viewing the page where the experiment runs. The goal defines when a conversion occurs.

Conversion

A conversion happens when a visitor reaches the defined goal.

Winner

The winner of an experiment is the variant that produces the highest number of conversions.

How Experiments Work

• Within a single experiment:
  • Only one goal can be defined.
  • You can test between 2 and 26 variants of a single component.
  • Any number of components on the same page can be tested.
  • You cannot add more variants or components after the experiment starts.
• On a single page:
  • Multiple experiments can run simultaneously, each with a different goal.
  • You can run an experiment with the same goal as another experiment on a different page.
• Variant distribution:
  • At the start of an experiment, each variant is served to visitors with equal probability.
  • As the experiment progresses and conversion data accumulates, the system increasingly favors the variant with the highest conversion rate.