A/B tests allow you to compare two versions of the same product and monitor the effectiveness of small changes in design. If you own a website, an app, or any other digital product, this method can help you test the changes that convert more and find out how can they improve the user experience through the analysis of conversion statistics and usability of the different versions. If the new version works best, you define it as winner and it starts to be displayed for 100% of the audience.
Big companies like Facebook, Google and Microsoft often perform A / B testing to improve their products. You probably should have already participated in one of these tests while surfing the internet and did not even notice.
An A/B test conducted by Facebook in 2014 indicated that when users opened the social network application for iOS and were surprised with a personalized loading animation on Facebook (left), this meant that they blamed the application by the delay in loading . In a B version of the app, presented to only a few users, an iOS native circular spinner (right) was displayed instead of Facebook’s default. Experience has shown that with this, users were less likely to blame the Facebook app and put the blame of the slowness on the OS.
— deeje (@deeje) 31 de janeiro de 2014
What to test?
Or maybe the site is trying to convince other visitors to register? All these questions can be answered, one by one, testing the elements properly.
Some examples of what can be tested on a website:
- Headline of the page
- Call-to-action (buttons for conversion)
- Images or videos
- Product Overview
- Size and form fields
- Reliability indicators (testimonials, certificates, etc.)
- Advanced Testing may include pricing structures, sales promotions, navigation, menus and much more.
Using Google Analytics to perform the tests
Since 2012, A/B testing can be done via Google Analytics. There is a tab within the tool to help you in the test creation process.
Google Analytics measures the effectiveness of each version of the page, and, with an advanced engine statistics, determines the most effective version.
The tool also provides a different model of tests, allowing to analyze two or more versions of the same page (multivariate tests A/B/N). They may be small changes, or you can test completely modified pages for specific groups of users from different URLs. See what you can do:
- Compare two or more web pages or application screens
- Define what percentage of your users will be included in the test
- Choose what purpose to test
- Receive updates by email about your experience running
A/B tests and multivariate tests are excellent resources because they offer real feedback from the market, measured accurately. It is not a simple research in which someone can answer one thing and do another one; but they are consolidated facts. The different versions are distributed randomly in the same duration without risk of external factors influencing the conversion rate, making it a much more reliable result.
Do experiments like this in your company’s products. The results are very satisfactory for you and especially for users who, consequently, will receive an improved version of the product and obtain a better experience.