The eBiz Blog

When it comes to optimizing your website to make it as effective as possible to your target market, A/B testing is invaluable. Instead of guessing your target audience’s preferences, you can use split testing to determine exactly which version works better and make better decisions.

But to get the most from your tests, you need to design and implement them effectively.

Manage Your Sample Size

Anyone who has studied statistics knows that in order for the results of a study to be considered valid, the sample needs to be large enough in order to be statistically significant. But when it comes to A/B testing, a large sample isn’t always the best way to determine a clear winner. Many businesses automatically use a 50/50 split, with half of the site visitors getting one version of the page and half getting the other. However, if you don’t receive a huge amount of traffic to start with, using such a large sample size could result in lost opportunities.

For that reason, in the early stages of A/B testing, you may be better off doing a smaller sample, such as an 80/20 or 70/30 split. That’s still a large enough sample to allow you to gauge the impact of the change, without negatively affecting the results you’re seeking.

Test One Element at a Time

One mistake that some new site owners make when using A/B testing is to throw the baby out with the bathwater. In other words, they change everything, testing two vastly different versions of a particular page. The results of such a test can’t be considered valid, as there is no real way to determine which element is the one that makes a difference.

Instead, test one element at a time. Change the font, the color scheme, the call to action or the tone of the copy, not everything at once. That way, you can determine the changes that make the most difference, and keep the elements that are working.

Be Bold With Your Changes

Because so many tests involve half of the site visitors, many site owners are reluctant to make significant changes for fear of turning off those visitors. However, in order to get a true picture of what’s working, you need to be bold with your changes. A/B tests need to be simple — if you want to do more changes, you may need to invest in the more complicated multivariate testing — but make the change obvious. Do more people click on a button that says “Buy Now,” or are they more likely to click “Get More Information?”

Perform Multiple Tests

In general, A/B tests are relatively easy to implement and assess, so there shouldn’t be anything holding you back from running multiple tests. Before you make any permanent changes, perform several tests, increasing the sample size each time. It’s possible that the results from the first test could be a fluke, or you didn’t get enough traffic during that period to create viable results. Either way, perform at least two or three tests before making changes permanent, to ensure that you are optimizing your site.

A/B testing has become a vital part of any Web optimization strategy. Before you embark on a full redesign, or even officially launch your site, test various elements to ensure it’s working exactly the way it needs to.