An Introduction to the Wonderful World of A/B Testing

The wonderful world of A/B Testing

|

08. December 2016

 | 

In part one of our series about A/B testing we will be explaining what A/B Testing is and how it can help improve the performance of your site. In addition we will outline the different types of optimization tests and explain them with use cases.

What is A/B Testing?

Website Optimization and testing has become a major topic in digital marketing. Gone are the days where website changes had to be built, deployed and then compared to historical data to discover if they were improvements or not.

Website testing can now be done by comparing the performance of two live versions of the same webpages in near real time. You can test two or more variations of your page against each other by splitting your web traffic between them and discover which is the most successful.

When A/B Testing tools were first starting to hit the market they functioned in one of two ways.

  1. Two versions (an original and a variation) of the same page would be deployed to two different URLS. i.e. example.com/a and example.com/b. The tool would then compare the performance of these versions to find the most successful.
  2. The tool would enable the two versions of the page to be shown on a single URL but these variations would both have to be hard coded on to the page.

Modern testing tools have made things easier for marketers and content professionals. With just one snippet of JavaScript on the page these tools are up and running. Then variations can be built using a point-and-click visual builder that allows for reordering of page elements as well as changes to color, images and text.

Why Test?

Modern website testing tools allow you to experiment with changes to your site without having to invest vast development time and money. They do this without you having to change any website code or take up valuable and expensive developer resources. You can quickly create and deploy various changes to the site and test which performs the best as well as customize the experience for different visitors.  This means you can test constantly resulting in tight and meaningful optimizations of your landing and content pages.

The bottom line here is that you can improve the performance of your website without having to make any code changes at all. Then only the successful changes need go into the development process.

In addition, these modern tools allow for hyper personalization of your website. Ensuring you present the most captivating message for each website visitor.

Types of Optimization Tests

There are several common types of optimization tests to help you improve your site. Each one is best used in a different scenario.

A/B Tests

A/B tests are the original website test. At first these tests would compare your existing page (A) to the new page you wanted to test (B).

Today A/B tests usually compare changes to a single element (or a group of elements) on a page.

Use Case:

Will a red or blue Call-To-Action button draw more clicks than your current green button?

Split Tests

Split tests (or sometimes split URL tests) are more akin to the A/B tests that could be conducted with the first wave of testing tools. They deploy different versions of an entire page to two different URLs. The testing tool will then split your traffic on the original URL between the two versions. They are best employed when comparing totally different page designs that are aimed at driving the same action.

Use Case:

You have two possible versions of a lead generation page. Do you use the one that focuses on pricing plans and packages? Or the one that focuses on product features? You can run a test to find which version leads to the most contact forms being submitted. 

Multivariate Tests

Multivariate tests are slightly more advanced tests. They compare variations in multiple elements. The tool will combine sets of these variations until it finds the optimum combination to achieve the page’s goal.

Use Case:

You have a page with a CTA button as its primary goal. There are a number of elements on the page. The button, a headline, a large image banner and some copy. Let’s assume we could deploy any of three different:

  • button colors
  • texts for the button
  • blocks of copy
  • headlines
  • image banners

In this situation it’s not just which individual variation will work best but in which combination do they all work best. The multivariate test will present every possible combination of all these element variations to users until it finds the most effective one.

Personalization

Personalization is a common feature of optimization tools that has come to the fore only in the last few years. It is less a testing feature and more a manner of customizing your website for specific groups of users. You can change page elements based on which user is seeing the site. Think about the impact you could make just greeting a customer in their own language. Why say ‘Willkommen’ to someone from Japan or ‘Kon'nichiwa’ when ‘Hello’ would be more appropriate?

Use Case 1:

Your restaurant is open for both lunch and dinner. If someone visits your site at ten in the morning you could feature your lunch menu but if they are visiting at four in the afternoon then display your dinner menu instead. Doing this makes sure your customers have the right information in front of them when they need it.

Use Case 2:

You run an online store that ships clothes all over the world and its mid-December. If someone visits from Zurich you should be featuring sweaters and jackets. However, what if they are in Sydney? Then it is time to feature summer dresses and swimwear. With personalization the same landing page can do both. Which will likely lead to more sales for you.

Getting Started

Now that you know what A/B testing is and are sold on its benefits it is time to get started with your own testing. In our next post we will show you how to plan your testing strategy as well as give you five strategies to make sure you win with A/B Testing.

Add comment