What is A/B testing?

by Startacus Admin

Over the course of your Startup life, you will hear many startup terms which might leave you scratching your head, or nodding along with a conversation, without perhaps really even knowing what is being discussed.  A/B testing might well be one of these terms.  

Like all such pieces of technical jargon, this one isn’t nearly as complicated as it first appears.

A/B Testing

The term A/B testing is used in a number of fields, but most commonly it is used to describe the process of a comparison of 2 different versions of the same web page, to ascertain which one performs more favourably towards a set goal, or goals.

This process is completed quite simply, by presenting the 2 variants of the web page, to largely comparable visitors at the same time.  The behaviour of each visitor is then analysed to determine which of the versions encouraged the preferred response.  

Why is A/B testing done?

All web pages have a goal, and A/B Testing is used to measure how the various elements of a web page conspire to drive visitors towards that goal.  Generally speaking, the range of goals that web pages have are fairly limited; most are intended to either sell products / services, encourage advertisement clicks, sign-ups to free / paid services such as newsletters & access to resources.  There are of course several other web page goals, but these are the most common.

How well a web page manages to drive its visitors towards these goals is known as its conversion rate; A/B testing is an attempt to implement changes that will improve the conversion rate.

What can be A/B tested?

Pretty much any element of a website can be A/B tested; the general rule is, that if it impacts (or has the potential to impact) the behaviour of a visitor to a site, it can be A/B tested. This includes everything from headers and paragraphs of text, to image placement and social media icons. 

How is A/B testing carried out?

The majority of A/B testing can be carried out fairly easily, given a basic level of technical expertise. It can be a time-consuming process, especially for web pages with a low level of traffic, so it is important to create a structured plan at the very beginning.  

If you are a methodical person, you will enjoy A/B testing because it requires a meticulous and scientific approach to ensure the best and most accurate results.  Like all scientifically valid investigations, you will need to follow a certain procedure.   

The general steps that you will need to follow might look something like this.

1. Come up with a question / issue that you would like to find the answer to, for example ‘Why are fewer people than I would expect signing up to my newsletter?’

2. Examine and assess the behaviour of visitors to your site, particularly with regards to newsletter sign-ups.

3. Decide upon your testable variant, and come-up with your hypothesis,  for e.g. ‘Moving the newsletter Icon closer to the header will increase the rate of sign-ups’

4. Put it to the test - Create 2 versions of the page you would like to test and compare the behaviour of visitors to each, in relation to newsletter sign-ups.  If there is a discernible difference in newsletter sign-ups on either of the versions, you will know that the conversion rate with this web page is higher than with the other.

That’s pretty much all the basic info that you need to know about A/B testing. 

• #a/b testing
• #development
• #startup term
• #technical
• #web

Published on: 17th June 2015