AB testing

How we improve conversions with A/B tests

21.11.2019

A/B testing is around for a long time, yet a surprisingly small amount of companies do it. People will spend days deciding on what the main front page banner should say, but when they do decide, they never bother to check whether they were right.

We say test it! We do and we get great results. 

What is A/B testing

A/B testing is a practice of displaying different things to separate groups of users and measuring what works better. A classic example would be to display a red call-to-action button to half of the website visitors and a blue one to the other half. We would then measure which colour gets clicked more often. 

The colour of a button is, of course, a simplistic example. In real situations, we can measure various things, from images, text, designs, up to whole layouts. This allows us to gradually improve the website and be sure we are doing the right things. 

Why do A/B test?

Test away and improve conversions 

Small changes to websites matter. A change of an image, a change of a call to action, adding or removing links to contact  - all these things can have an impact on what your users do and eventually on conversion. Without A/B testing it is extremely difficult, if not impossible, to decide whether the changes we introduce make sens and also - which changes to introduce. Also, there is no incentive to play around with these smaller elements if we do not see the result.

A/B testing allows you to implement the changes which increase conversions and test our various hypothesis. It is much easier to change the main call to action texts, change designs and try out new things if we can easily check whether these work. 

Avoid making costly errors

A/B testing, it is not only about improvement, but it is also about not damaging what we already have.

That’s right - A/B testing is great when you are building a new website, but it is EVEN MORE IMPORTANT when you already have a website that works. If your corporate website drives sales (contact forms are filled, orders are submitted) and any changes you make to it, put you at risk of destroying this value which is already there. If you do not measure the results of your actions, you will not even know you are breaking your website.

A/B test results compound 

A single A/B test rarely gives a significant change but combined, they can have a really strong impact. Let's look at an example.

For our client - TrainingRealm.com - we recently performed 2 simple tests, which allowed them to increase performance by a full 31%. The tests were really simple:

1. We changed the heading of the home page banner

TrainingRealm Ab test

Initially, the banner text said: ‘Training Realm - the ultimate training instructor’. We suggested that a more explicit heading might be better. We suggested to change it to: ‘TRAINING REALM 100 PUSH-UPS, 50 PULL-UPS 300 SIT-UPS’ what helps users who come from satellite training websites quickly understand where they are.
Result: 18% increase in registrations.

2. Add testimonials on the subscription purchase screen

AB test 2

We expanded the subscription management page by adding user testimonials with faces.
Result: 11% increase in conversions to paid subscribers

Overall these 2 changes caused a 31% increase in the number of subscriptions. 

Any further changes we make to the app will have a compounding effect increasing the subscription rate even further. If we find a change which increases conversion by 5%, the overall improvement of the 3 changes will be 37.5% (you can see how it compounds)

What is also important - we will never introduce a change that reduces subscriptions. We will spot these early and remove them quickly.

Try out A/B tests on your website!

 

Looking for web development experts?
Check our services

All services
Drupal
Symfony
React
PHP
DevOps
Nodejs
Design/UX