Build Experimentation-in-Personalization campaigns

  • Updated
You must have both Optimizely Personalization and Optimizely Web Experimentation for this feature.

In Optimizely Web Experimentation, the Experimentation-in-Personalization feature lets you test and personalize experiences at the same time. You can A/B test your personalized experiences to customers, create two versions of an experience in a campaign, and compare how well they perform.

Optimizely Personalization lets you deliver tailored experiences to different audiences. Use Experimentation-in-Personalization to test multiple variations for a single audience in your personalization campaign.

Visitors who meet the audience conditions for a certain experience are randomly assigned to one of the different variations of that experience.

For example, you may personalize the messaging on your homepage for two audiences, such as Luxury Travelers and 5x Shoppers. You know the content you want visitors that qualify as 5x Shoppers to see. However, according to your research, visitors who qualify as Luxury Travelers prefer a combination of luxury goods and a sense of adventure. 

You create and test two different variations (luxury goods and adventure) for the Luxury Travelers audience in your personalization campaign. 5x Shoppers see the 5x Shoppers experience. Luxury Travelers are randomly bucketed between the luxury goods and the adventure variations.

The Optimizely Experimentation Results page shows which variation performs better and how well each audience in the campaign performs overall.

Add a variation to a personalized experience

Start by creating a personalization campaign.


Each audience attached to the campaign contains a single variation and a holdback. Add more variations to experiment with multiple variations on a personalized experience for a single audience.

  1. Click Add Variation.
  2. Name the variation, then click Save.
  3. Click Edit to modify a variation in the Visual Editor.


When all experiences and variations are ready, click Publish to deliver them to visitors.

Visitors who meet the audience conditions are randomly grouped into the variations. On the results page, you can compare how visitors respond to the different variations and track results for that audience as a whole.

Traffic allocation

Traffic is automatically split evenly between variations.


Visitors who qualify for the audience are randomly bucketed into one of the variations or the holdback.

You can change the traffic allocation for the variations.

  1. Click More (...) > Settings.


  2. Edit the traffic distribution percentages for the variations, then click Save Experience.
Total traffic distribution must equal 100%. If it adds up to more or less, a warning message displays, prompting you to adjust your distribution.

Change the holdback

The holdback in a campaign lets you track the performance of personalized experiences you deliver to each audience compared to a generic experience.

Each campaign has one holdback that remains the same regardless of how many experiences and variations you add.

See how to change your holdback.

How many variations can be added?

There is no limit to the number of variations you can add for each audience in a campaign. You can prioritize audiences so that if a single visitor qualifies for multiple audiences in Optimizely Personalization, they see only one experience by default.