Guest Post: Setting Up A/B Testing Correctly

A/B testing improves website conversion rate

Image courtesy of Mil8

My final post in my own series on the new testing approach to website design still has yet to move from my ‘to do’ list and onto the blog. So to fill the void, here’s a guest post from Corbo and how to run A/B tests correctly for higher conversions and more sales. 

If you’re after maximizing your profits to the fullest extent, employing a/b testing is a great way to gauge the efficiency of your business strategy. Here’s how you do it simply and quickly:

First of all, you’ll have to decide what factors you want to test on. This can be anything from the popularity of one product over the other to the viability of sales of one branch against the next. After which you’ll have to set up the test to ensure accuracy and efficiency when it comes time to calculate the results. There are a slew of various testing programs that you can use to ensure the precision of the task you’re about to undertake.

One of the most popular and user-friendly tools available for a/b testing is the Visual Website Optimizer, a very easy-to-use tool with lots of extras that make testing quick and easy. If you’re a bit savvy with ins and outs of HTML and Javascript, opting for Google’s very own web optimizing tool should be a breeze for you. Of course, there are a myriad of other tools that you can choose from, with Vertser, SiteSpect, and Test & Target by Omniture running the lead. It should be understood that the setup of the whole thing is the same for practically every tool out there, but it is the procedures and how they’re done that decide the final results.

Decide on the setup of the test through the following methods:

Element Replacement

Before your page loads, you can create different variations of one element such as a button or icon via your testing tool with the use of HTML. Once the testing goes live or once you’ve gone to the beta stage these various replacements or element alternatives will be displayed randomly on a page prior to being shown to visitors. This change allows you to decide which aesthetic suits your site best over others without having to create multiple dummy sites.

Redirecting to an alternative page

If you do decide to test a whole page, you can choose to create a new page and upload it to your website as a variant which automatically functions by redirecting a visitor to the alternative site. If, for example, you begin with a http://www.example.com/index.html, you can create an alternative via slight modifications and make it out as http://www.example.com/index1.html, and select an automatic redirection from the original page to the alternative. This allows you to create multiple variants of a single site to test its look and dynamism.

Setting Up a Test:

  1. You have to choose a page (or a number of pages) to test.
  2. Thoroughly define the elements that need to be tested.
  3. In order to facilitate faster and more efficient tracking of results, add tags to the original page and every alternative thereafter.
  4. Enter the website’s URLs into the tools testing window prior to every test.
  5. To ensure consistency, make sure that the A and B variants of the primary site is tested at exactly the same time, ensuring that the visitors see only a single variant regardless of the number of visits they attempt.
  6. Check the results of your tests in the dialogue box after every run and record the results for later review.

After you’ve received enough traffic to ensure thorough consistency, review the results and incorporate the dynamics or aesthetics of the winning test site into your main website or campaign strategy. As a final note, the following should always be remembered:

Test long enough – this is to ensure that you get sufficient amount of traffic to allow for statistically significant results. Don’t give up during the initial test run, as it is a given that you will have very few clicks. Wait until it reaches a quantitative amount on both sides prior to calling the test through.

Surprise isn’t good – not for old clients that is. Always remember to test out your sites on newbie visitors and always remain consistent with the layout during the initial visits (yes, include repeats in the test). To avoid blunders, always perfect your code and incorporate provisions.

Consistency is key – in order to ascertain whether your a/b test is thoroughly reliable, always maintain consistency by presenting the same element on a single test page per test run.

In Conclusion

A/B tests aren’t really all that difficult, and with the proper tools that can help you implement reliable and efficient a/b testing, you can now try out your site’s ultimate business potential.

Ruben Corbo is a freelance writer and writes for a number of online marketing websites including those that help online businesses improve A/B testing techniques to increase sales conversions. When Ruben is not writing, he’s producing or composing music for short films or other visual arts.

No comments

Leave a Reply