Split testing or a/b testing is one of the easiest methods of increasing conversions and adding a little more r to your roi. All sorts of savvy businesses and multi-nationals use these killer techniques to gain massive insights on visitor behaviour and conversion rates.

The evolution of the internet, ever-improving software and mobile apps have made split testing more accessible and practical

How would a one percent increase in conversions affect your company? What about a five percent increase in conversions or even a ten percent boost in conversions?

Making a positive impression on your conversion rate is priceless and it’s not as complicated as you think. After all, what could be better than figuring out the best online promotional and marketing strategies for your business? Using split testing, we have increased conversion rates for our clients from 22 percent to 40 percent just by changing the background of certain important pages on their websites.

Split testing allows you to test nearly everything on your website or other marketing materials including headlines, buttons, banners, calls to action (CTA) and so on. Finding an element that performs better than the alternatives will make much better use of your online traffic or marketing material. Split testing has existed in some form or another for as long as there have been merchants putting up signs and displaying their wares.

So while the concepts themselves are not entirely new, the evolution of the internet, ever-improving software and mobile apps have made split testing more accessible and practical. By the year 2000, tech giant Google was running split tests on everything including the number of search results to display on their search results pages. Today Google’s split tests number well into the tens of thousands.

Up to now we’ve discussed just one variant of split testing, A/B Testing, but there are a number of different split testing methodologies you can adopt to improve your performance.

The most popular methods include:

  • A/B Testing
  • Multivariate Testing
  • The Taguchi Method
  • Web Usability

Multivariate testing

This method of testing allows you to carry out many tests concurrently. Testing which headline to use, text, images, prices, offers, and buttons can be conducted in a single test. In a multivariate split test, visitors see different combinations of these elements at the same time. The results show which elements perform best and this data provides the info required to create a page featuring only the highest converting elements.

The taguchi method

Named after Mr. Genichi Taguchi, a Japanese engineer and statistician, this method has been applied to everything from automotive R&D to advertising.

Taguchi’s multivariate (two or more variable quantities) testing tools can gather results based on fewer tests than classic A/B testing, but the jury is still divided on Taguchi and the validity of results.

Split testing software packages worth checking out


Start small and run your tests on significant segments of your audience to produce valid results. With a little practice, you’ll be boosting your own conversions in no time.

Web usability

A technique used to evaluate a website by testing it on users. This is a great way to improve the user experience on your website. It highlights sticky areas that could be improved. In this type of web testing the idea is to create a scenario wherein a person performs a list of tasks that a first time website visitor is likely to perform. These results are in the form of heat maps showing the most clicked or scrolled areas of the site.

Focus on things that are most likely to have a huge impact

With so many options for testing and improving the elements on your web site it can be easy to go overboard. As with many things in life, moderation is key. There’s no need to start testing every element on your website immediately to see an improvement. In fact, this can be counter-productive. Instead, focus on the things that are most likely to have a big impact. Start with the simplest elements you can and work your way through all over time.

How one simple change improved our conversion rate by over 40 percent

Here’s an example of a split test that we conducted for one of our clients. This particular client releases new products on a regular basis and are always keen on us running a few tests for them to determine areas where we can improve their conversions. So this particular test was run across four of their most recent product launches and the results were quite impressive. We picked an element of their website that is usually overlooked. In an attempt to boost conversions, we decided to test three versions of one of their main content pages with three different background images.

We decided to test three versions of one of their main content pages.”

Graph image here….

This simple experiment turned out to have a massive impact

Here are the parameters of the split test: Essentially, all the content on the three pages remained exactly the same. In fact, all other elements of the web page were identical. The only difference was the background image being used.

During the first product launch the test pages were created and added to our split testing software. The split testing software evenly distributed the client’s website visitors to one of the three pages with a different background image. At the end of the split test, the page with the black background emerged as the strongest (see graph above). Armed with this information, we simply disabled the pages with poor performing background images and let the winning page be shown to all our client’s website visitors.

Split testing helps you get more bang for your marketing and advertising bucks

Running a few simple split tests with a sample group allows you to identify the highest converting combination of
web page elements on your pages. With your split test data you can tweak your web pages and/or offers and place all the elements on the page in their highest converting configurations before you start pushing large volumes of traffic to your web pages.