Website A / B Testing – Why and How?

As a company’s CEO or head of marketing you certainly know the power of “You can always do better” attitude. Whether it’s increasing the productivity of the company, selling more products, creating a new website or launching a new TV commercial. You just know things always can be improved. Still at one point decision should be made.

In this post we’re going to look at how to make your website convert better. How to get more qualified leads and orders using the power of A/B testing. This method is very powerful if you know how to use it. Our goal is to find out those elements on the website that converts the visitors to clients. At the end of the day it doesn’t matter how many visitors you have on your website. It’s sales, orders, subscriptions that matters!

Website A / B Testing - Why and How?

Main topics we will cover:

  • What is A/B testing?
  • Why A/B testing is important?
  • How A/B testing should be done?

What is A/B testing?

A/B testing is a process that helps you to find the best performing website layout. Users will be showed different versions of your website. For example, 50% of visitors will see one version and others will see another version of the website. By using this method you will notice soon the details that most influence your site visitor. By continuously creating new tests you may increase your website conversion rate a big time.

A/B testing is like working out at gym.  If you do it once, you probably do not notice a big difference. If you repeat it for several times, you start to see the progress. If you stop training the results remain for a while but start to decrease soon. To get the best results you have to focus and continuously work on it. The same goes with the website. First you have to figure out the elements that most probably will change the visitors into customers. You can test all different elements of your website: headline, content, picture, testimonials, Call-To-Action (CTA) buttons / size / text, videos etc.

Why A/B testing is important

While creating a new website or landing page you cannot be sure that you have right headline, picture, content or CTA. You have to test to find out the best performing one for your visitors. You have to figure out which information or element works best for your visitors. What makes them to decide about buying your product. It doesn’t mean you have to create two different websites, you might want to change few details. With different A/B testing programs you can easily create different versions of your website.

How A/B testing should be done

At first you have to understand your website visitors profile: age, gender, education, interests. How they use your website, what are the top visited pages, average time on site, where are they clicking etc. You cannot assume anything or trust anybody besides your own website data! You are not building a website for yourself but for your clients. Put yourself into customers shoes. Start from that. If you test different elements and make right decisions you definitely increase the number of website conversions.

1. Collecting the data

The first thing you should do is to analyse the data in Google Analytics. How people reach to your site? What is the average time on site? Which pages are most visited? What are the common things most of the buyers do on your website? What kind of information they need before a purchase?

If you don’t know how your users behave on your site then check out Real people will test your website (or app, or wireframe etc) by your guidelines/tasks and give feedback.  This is really good stuff!

All the sessions will be recorded and you could analyse them and make your own decisions. You could get answers to your questions like: did the visitors find the needed information easily? Why they abandoned the shopping cart? What kind of information and how are they looking for. You should get an overall overview if you have 4 upto 7 testers. Then you have enough data to analyse and some common patterns will show up. For sure you can take 50 or more testers as well but this needs more time and resources which most of new A/B testers don’t have.

2. Creating hypothesis

Once you have collected the date you need to set up some hypothesis that what could be improved on your website. What is wrong and how it possible could be made better. Example. Hypothesis 1: Too few people register to my client program because of the complexity of the sign up form which has 10 fields and 8 out of them are complimentary. Solution 1: Replace the form with 4 fields and only 2 of them are complimentary.

Hypothesis 2:  Product “ABC” gets very few sales because it has only 1 small picture which cannot be zoomed and there is very short description about the product. Solution 2: By adding 4 new quality pictures with the option to zoom and adding additional text to product description visitor will get more information and are more willing to buy a product.

3. A/B testing – Control, Variation A, Variation B…

Now as we have figured out the possible problems on website and made hypothesis we need to test those. We don’t necessary need a programmer or designer for that. We need some programs that could do the work for us with little HTML, CSS knowledge.

One of the easiest A/B testing tool is VWO. With it you can create 2 or more variations from your website. You only have to add VWO script to your website and you will be able to make modifications from VWO.

Another good landing page platform with includes A/B testing possibilities is Unbounce. You can easily create 2 different landing pages and run those at the same time to get data which one works the best.

Control vs Variation 1

It’s important to understand some A/B testing specific terms like Control, Variation A and Variation B. Control is the main (current) version of your website. All the other version will compete with this: like Variation A and Variation B. The best way to start is to make one change at time and see the results – Traditional A/B testing. Then it’s easier to make decisions if there is only one elements that will make the change. One Hypothesis = One Test. Still sometimes there are many elements that make the difference. For that we use Multivariate testing. Running this test we test many elements in the same time. For example we would like to know which combination works better Gallery, red CTA and Short Form or HD video and long form.

4. Analysing the results of A/B testing and retest.

To make the right decisions the A/B tests should run 4-12 weeks depending of the site traffic and hypothesis. Some results might be seen earlier as well but you cannot be sure once you haven’t ran the tests long enough. Then you are able to see which Variation (or Control) gets the best results. Stop the lowest performing Variations and create a new Variations to beat the previous winner. This is how you always find the new ways to make you site better and better.


A/B testing is definitely something that make your website better and increases it’s conversion rate. Dig into your Google Analytics and User Testing data and create a buyer profile. Try to understand what converts the visitors into customers. Create hypothesis and ran A/B tests continuously. Your website can always convert more! It doesn’t matter who win in short term, champions know how to win in long term!

Facebook Twitter Google+ LinkedIn Pinterest 

Add comment

Get Your Copy of Our Latest Marketing Tips!