A/B Testing: Taking Your Website From Good to Great

A/B Testing: Taking Your Website From Good to Great

By: Shelly Kramer
March 25, 2011

A/B Testing: It Makes What You Do Better

Something you should be considering for your marketing efforts, especially as they relate to your website: A/B testing. Sounds fancy, but it’s really just a little testing of your concepts – and some math. And, of course, spiffy tools.

Oh, and did I mention that a little time spent testing can result in impressively different results? Like perhaps selling more things to more people. Or getting more leads or conversions from your website. Little things like that – well, they make me happy.

A/B testing is really quite simple. You have two versions of something and a metric that defines success. You experiment, measure, and typically use the one that had the best results. Duh, right?

For A/B testing on your website, you can use a program that directs visitors to one of two different versions of your site and then tracks the data for each of the versions, so you can easily see which version performs better.

One of my favorites sites for this kind of geeky goodness is Whichtestwon, which has an A/B testing “test of the week” post where you can guess which version of a site performed better. You can then vote, see if you were right or wrong, and then see what other people voted. You can even tweet that you got it right (I did today!) or wrong (a gutsy move).

While I geek out on “testing my gut” like this (and spent entirely too much time checking out that site), the concept of A/B testing is not only fascinating, it is not a new concept. Marketing and Ad agencies have been running numbers and charting results of every campaign for as long as there have been ads. Knowing that sales go up when a middle-aged brunette (like me) sells a mop versus a voluptuous Barbie impersonator (like Gini Dietrich) (I’m testing her reputation management skills here) is part of understanding your target demographic.

And for this kind of traditional market research, it often costs all kinds of money and takes time and energy and resources. But today’s A/B testing is automated. And very slick. More importantly, it can help you determine some pretty important factors in your website.

The way it works:

There are a few options for A/B testing your site. Google’s Website Optimizer is one of them – and it’s free. Some others include Performable or Optimizely, which are both paid tools.

Once you have chosen your provider, you create two different versions of the site you wish to test. For example one might read “Save $100 on Wine of the Month Club” while the other may read “12 bottles for only $69.99.” The product costs $69.99 regardless, but one highlights savings, and one emphasizes how many bottles you get for how little.

When a visitor comes to the site, they are directed to one or the other version of the page and the metrics of the visits are recorded and charted and analyzed so you can determine which approach worked better to increase sales (or whatever your goal is).

So after a time you find out if version A or version B performed better and you can adjust other pages/sites to reflect your findings.

This kind of testing can also be applied to email campaigns and other forms of marketing as well. You can test which email subject line gets a higher CTR rate. Or you can test which opt in box form captures more emails. Testing can even be done on social marketing campaigns.

There are other tools, too. Ones that make my head swim at the deliciousness of all that data.  There are heat mapping tools like Crazy Egg, which lets you visualize your visitors’ behavior in some uber cool ways and Click Density, which helps you use real-time visitor data to refine your site’s data and performance. HubSpot’s another resource that has some great tools to help you improve your site performance and conversion rates.

The bottom line is, with A/B testing you can get some real, measurable data to hone your marketing campaigns, cut the fat and make things happen. And nothing can boost your marketing performance like good solid data.

In our house, it’s all about data. We design. We test. We gather and analyze the data. Tweak. Start cycle all over again.

What are the visitors to your website doing? And what are you doing to maximize results?