A/B testing for your website. Improve visitor interaction with your content.

A/B testing is one of the modern method of evaluating the performance of a layout and compare it with another layout. A/B testing also called as ‘Split testing’. Lots of internet marketers are using this proven method as one of their secret weapon. At the base, we generate two distinct layout and put them online using randomized script. Then script will show chosen layout to visitors. Using A/B testing, you could easily increase your website conversation rate.

We’re going to explain the very basics of the A/B testing. When we produce a new layout change, it’s not always good to switch it instantly. When designing a good UI, first thing to remember is. “You’re not the user” and next thing is “You’re the user”. Pretty confusing, ain’t it. We’re surely going to cover UI measurements with our upcoming posts.


1) A/B testing in a nutshell

Imagine a web developer/designer who needs to push his new layout or change in to an existing website. But your boss says ‘NO’, it’s not the design that we need. Your company goal is to increase the website conversation rate. You believe that your new change will allow users to interact with website more fluidly. But you’ve no solid measurements to prove your boss that he/she is wrong.

A/B testing - Two Faced

Let’s get back to your school science laboratory testing. Imagine an experiment that you’ve tested two dynamics and compare them to get the best result. Imagine an Acid reacting test to calcium carbonate. Small chunks will react more furiously rather than putting big calcium carbonate chunks. But without the experiment results, it’s not the scientific way to state that small chunks reacts more furiously than big chunks. Similar concept is used with the A/B testing. Here we’re testing two dynamics and calculating the best result.


2) How not to setup A/B testing (Split testing not to do method)

First of all, there is a Myth on A/B testing which is A/B testing is an equal for Cloaking. Cloaking is a black-hat SEO method that uses to gain quick ranking on search engines. What cloaking basically is, showing a different version for search engine Bot and showing entirely different version for users. Nowadays cloaking considered as an outdated method. Because bots are quiet intelligent to find the cloaking. However in Good ‘ol days, you can search for ‘Cheap USB Drives’ and end up in a ‘Viagra’ website.

A/B testing - Cloaking on user agent

As long as you doesn’t select based on the User Agent specified for Google or similar search engine, you’re good to go. This is how ‘Google’ describes the Cloaking.

Cloaking—showing one set of content to humans, and a different set to Googlebot—is against our Webmaster Guidelines, whether you’re running a test or not. Make sure that you’re not deciding whether to serve the test, or which content variant to serve, based on user-agent. An example of this would be always serving the original content when you see the user-agent “Googlebot.” Remember that infringing our Guidelines can get your site demoted or removed from Google search results—probably not the desired outcome of your test.

It can demote your website and probably you will end up in the last page of google or worse. If you’re going to use A/B testing to select the best conversation enabler, do not use mentioned cloaking techniques. But developer can use the User agent to find the operating system and produce a webpage based on that. We’ve used similar technique on older design of http://mozilla.lk

3) How to setup A/B testing (split testing to do method)

A/B testing is a fairly simple method to use at base level. First you place a script that randomly generates 1 or 0. Then you place your two layouts and directs users based on that random number. Here the probability of user direction per page is 50%.

A/B testing - How to do it

As on our demonstrative image, you’re putting your clients as a test subject. Client doesn’t know that s/he’s on a testing page but thinks s/he’s on the real webpage. From hereon webmaster can provide separate analytic variables for the website. Both pages must be measured and evaluated.


Based on this test subjects, you can easily select the ‘Variation B’ since it has the highest conversation rate. Following are the best website types that you can use A/B testing (split testing) to increase the conversation rate.

  • Web applications that offer trial sign-ups
  • E-Commerce websites with product purchase option
  • Blogs with advertising gaining

But don’t limit your imagination for these types. You can alter the way you’re testing and use A/B testing on any kind of website. To reach your ideal goal, future A/B testings can be conducted based on the selected variation.

On our future posts, we will explain technical setup with the A/B testing.


One thought