Using Web Analytics to Measure A/B Tests

 In Data-driven Marketing, digital marketing, Standard, Web Analytics

One of the most effective ways to increase the conversion of your web site is to rely on A/B or multi-variate tests to improve the performance of your top landing pages or poor performing pages. Tools like Google Website Optimizer have made testing both easy and affordable for the masses, providing both an easy interface and intuitive reporting on the performance of tests. Within such tools, you can define your test criteria; your conversion points and you’re up and running.

But what do you do when the default reports within the test tools fall short? Consider the example shown in the figure below, which is an A/B/C test (one control page and two test pages). The results show near identical results between test pages B & C, with pretty good performance improvement with respect to the control page (75.4% and 75.7%) respectively.

So clearly, the test has been successful at increasing conversions compared to the original page, but how does one decide between tests B & C? Of course, we could give the test more time to run, but with near identical conversion results inside Google Website Optimizer, it is clear that we need a more comprehensive way to measure the performance of the two test pages. Enter web analytics. In this case, we’ve used Google Analytics and in order to look at these two tests, we’ve created two advanced segments where the landing pages are test B and test C pages.

Figure above shows the site KPIs for the two test segments. We can see that although the conversion rates are near identical, Test C outperforms Test B when it comes to site engagement, with higher number of pages/visit, more time spent on site and lower bounce rate.

So even though one can rely on the default reports provided inside the testing tool, we recommend that you compliment it with reports inside your web analytics reports. The added reports let you not only see conversions, but also look at the data more holistically by relying on other metrics such as site and content engagement.

Recent Posts