Thursday, January 17, 2013

Let’s Get to the Basics… A/B Testing


 Congratulations!  You have yourself a website, you’re collecting website analytics and you’re one of the few lucky ones to already have site traffic.  So it sounds like you’re doing great, but there is a problem.  How is your site conversion?  Is it struggling?  Is it not performing as efficiently as you expected?  Don’t give up yet…there is hope.

It doesn’t matter if your site is an ecommerce or a lead generation, or just an educational site.  Your goal should be to engage the client and get them to perform your desired action.  Regardless of what that action is, you will get your site to the highest performance with A/B testing.  As a result of my research, I have been able to answer these questions:  What is A/B testing?  How to A/B test?  What programs help you A/B test?

What is A/B Testing?

Optimizely.com defines A/B testing as:

A/B testing is a simple way to test changes to your page against the current design and determine which ones produce positive results. It is a method to validate that any new design or change to an element on your webpage is improving your conversion rate before you make that change to your site code.”

Simply put, you take your existing landing/home page and create a carbon copy of the LP/HP and change one element.  Then you split your traffic up evenly and track the results of your Key Performance Indicators.  Once you have gathered all the information and can determine a winner, you then apply the winner of the A/B test to the permanent LP/HP.

How To A/B Test?

Start by selecting the element you would like to test.  If you are struggling to pick an element, you can use the site called www.whichtestwon.com to view examples of A/B tests that others have run and found success.  This can help spur ideas about what you should test.

Determine what your goals are or KPI’s…the winner of the A/B test will have to improve the KPI’s.  Examples of possible KPI’s are leads generated, click-through rate, increase in revenue, etc.…

You will then want to split the traffic to your site, typically by 50% to the two variations of your LP/HP.  You don’t always have to do a 50% by 50% split.  Uri Bar-Joseph, from 8 Rules of A/B Testing – The Art in Marketing Science suggests that you can even do a 95% vs. 5% split.  You will just need to figure out the winner based on the winning ratio.

This is not a one-time process.  You will want to continually test and test and test.  Always gathering data, it will never lie or go on a whim.  By continually performing A/B tests, you will get your site to its peak performance.

For a more in-depth look at the steps you can take for implementing A/B testing, I highly recommend 8 Rules of A/B Testing – The Art in Marketing Science.  This article details more insights and tricks to A/B testing.

What Programs Help You A/B Test?

Two of the more popular software to use for A/B testing are Adobe Test & Target powered by Omniture and Google Content Experiments (formerly Google Website Optimizer).  They are both powerful tools.  Test & Target is a very robust tool that will greatly assist you in improving your site, however it can be a bit pricey.  Google Content Experiments is more cost effective but is a bit less robust.  With both of these, you may need some programming knowledge. 

Here is a list of the top 45 programs that you can also use to assist you with A/B testing, presented by Which Test One.


References:


·      8 Rules of A/B Testing – The Art in Marketing Science, Uri Bar-Joseph, August 6, 2012,  Search Engine Watch, http://searchenginewatch.com/article/2196754/8-Rules-of-AB-Testing-The-Art-in-Marketing-Science

5 comments:

  1. Thanks for detailed explanation of A/B testing and some useful resources. Non IT person like me can get a basic understanding of this process.

    ReplyDelete
  2. Thanks for sharing! Very useful guide to learning how to test different website ideas.

    ReplyDelete
  3. I love me some A/B testing. Mucho thank you, sir.

    ReplyDelete
  4. How much do the A/B test software cost?

    ReplyDelete