Continuous A/B testing of Homepage recommendation strategies

Channel

Mobile Web, Web

Impacted Pages

Homepage

Experience Type

Recommendations

Implementation Effort

Low
Who:

One of Europe’s largest electronic retailers

Strategy:

To increase customer engagement on the homepage, the electronics retailer displayed product recommendations to improve product discovery and push users down the shopping funnel. Realizing that there is not one recommendation strategy that would work best, the only best practice is to test, the electronics retailer tested different sets of recommendation algorithms in order to find identify the optimal algorithm.

Continuous A/B testing of Homepage recommendation strategies Continuous A/B testing of Homepage recommendation strategies
Hypothesis:

Product recommendations can drive many different user actions. For example, for new site visitors, they facilitate product discovery, but for returning users, they can resurface previously viewed products. In order to find the best strategy for each audience segment, we can let machines chose the best performing strategy for the average user. Doing so will optimize performance increasing click-through rate and ensuring customers will always be shown relevant products.