A/B TESTING: Reducing Bounce Rate

- Summary

When I worked for Marketscale, a client contacted us because they were worried about the high bounce rate on their homepage. Especially because a lot of visitors left the site without even reading about what the company does or what services it provides. We used Hotjar to identify a potential pain point causing new visitors to leave the site. Using an A/B test we tested a version of the homepage without the pain point and improved the number of people who reached the bottom of the page and the number of people who visited other pages besides the homepage by around 30% both.

- Challenges

The first challenge we ran into was figuring out what was causing the high bounce rate. By using Hotjar we noticed that around 50% of the visitors who started scrolling down, quit halfway through a very long banner. This banner became our prime suspect of being the pain point.

Another challenge we had was on choosing a sample size and how long the test would last. We settled for letting the test run for a week, to take into account the changes in the number of visitors during the weekend. We split the visitors by half and sent each half to a different version of the homepage.

- Data Preparation

We noticed that most visitors were already clients and only visited the homepage to login into their accounts. These visitors were not going to scroll down or read what the company does. Using the Google Analytics API we filtered out any visits that included successful logins,

- Data Analysis

Our main target with this test was to increase the number of new visitors who read what the company did. To accomplish this we needed new visitors to scroll down, where the text explaining the services of the company can be found, and also visit other pages besides the homepage.

Ee tested a version of the homepage with a much shorter banner and a version with no changes made. The version with the shorter banner performed much better with the two metrics we set for this test.