Website Optimization 101: A/B Testing Method


You might wonder why a certain website emerges as the leading one in its niche when there are hundreds or even thousands of other sites on the web that are basically serving the same content. Aside from thinking of the reasons why, the more important question to consider here is how they managed to do it.

The answer is simple: certain sites are more popular because they’re offering something that the other sites don’t. It could be a certain feature, how the information is presented, or the general ease of usability of the site. From this, you can already conclude that it takes more than just stellar content to become the top site in your niche. A good website is the sum total of the components that comprise it—and in order to find the best combination to use, you’ve got to do some testing.

What Is A/B Testing?
A/B testing is a marketing testing method where users are presented with two versions of a website which will only have one element that varies between them. The variations are served simultaneously to site visitors in order to gauge which one is better. Webmasters usually come up with a criterion early on to help them determine which version can provide the desired results.

A/B testing is very similar to another testing method called multivariate testing. In fact, they only vary in the number of elements that can be tested out for each test run, as more than a single element can be modified for multivariate tests.

Why You Need A/B Testing
A/B testing has been helping hundreds of thousands of webmasters and web designers all over the world with website optimization and design. There’s only so much that you’ll be able to get from design guidebooks and manuals. To come up with a really good design, you’ll also need customer input because the site is intended to serve them, after all.
Surveys are very tedious to conduct, not to mention a huge hassle to process. Getting a large enough sample size of users to fill out your survey could also prove to be a challenge. A/B tests let you forego these hassles by presenting a hands-on survey of sorts to your site visitors, where their actual and unbiased reactions to your content will prevail.


A/B Testing: An Example
A certain online clothing retailer is having a sale. To prepare for the event, the webmaster tests out two versions of the sale announcement. The first is a huge banner with images of sample items, along with the word “Sale!” The other banner only highlights one object from the sale and contains the promo duration as well as other information about the sale.

Both versions of the banner will be served to site visitors randomly. When the test is over, the results are brought up, and the one that brought in the largest number of sales is deemed as the better version.

Quick Guide to A/B Testing
Here’s a quick run-through of the entire testing process:
Content Assessment. The first thing you should do is look over all the pages on your website. List down which elements can be improved or modified, and rank them according to priority on your list.

Creation of Test Elements. Choose one component from the list to test out, and create the different versions to be used in the A/B test. Remember to make each version distinguishable from the other to get better results.

A/B Testing. When you’re ready, you can now run the test. Running these tests is made easier with the availability of tools and software, like Google’s Website Optimizer, which is available for free. If you prefer, you can also hire a third-party service or firm who can run the tests for you.

Interpretation of Results. After the test has run its course, it’s time to look at the results and determine which version gave better results. Implement the change on your actual website, and move on to the next item on your list.

Subscribe to my weekly update newsletter.

* indicates required
Are you a human? Check the box below.
Email Format


  1. Gault Boutique says

    nice read. I thought A/B testing is only applicable for PPC. It would also be useful in designs and for other purposes.