Testing methods in web analytics

So you’ve managed to get a good amount of traffic to your site. Search engine traffic, direct referrers through link building and probably a fair amount of traffic coming from any ads you might be running on other sites or search engines. After all the investment (both time and money) you’ve made to get this traffic the next step is to make sure your time and money are well spent by ensuring your website visitors perform the actions you want them to take. Your initiatives to improve conversion rates can be extremely rewarding in terms of ROI and simple changes on your site may have a significant impact on the number of times your set goals are reached.

The backbone of improving conversion rates is your testing system. It’s extremely important to first define how you’ll be tracking, analyzing and logging the changes you apply.

Tracking

The basics of tracking for optimization can easily be done with Google Analytics, which will provide you with the right tools to start testing website changes. Should you or your organization step up your conversion optimization initiatives you’ll probably need something more advanced (and costly).

The analysis

The analysis consists of two parts:

1. Know your site inside out. Know which your key pages are in the conversion funnel – Which pages do I want visitors to see and which pages to I want them to go to. If you offer search functionality know how it works, if your site offers a lot of menu options know the structure of your navigation, etc. Consult your web analytics reporting tool and study navigational paths and site overlays. Where do visitors enter my site? Which pages are popular? On which pages do I lose most visitors? On which pages of my conversion funnel do visitors exit most (bottleneck)?

2. Develop a list of initiatives. If you’ve been working with the same layout and structure for a while you might already have some ideas for improvement. There are quite a few sources of information out there of website best practices and usability and I advise you to read up on these subjects to be able to select tried and tested ideas that may benefit your particular site. There are loads of changes you could test so it’s a good idea to classify changes by priority. You’ll obviously want to start with those changes that have the highest potential of improvement. Limit your priorities to three classifications though, to keep a good overview.

Logging your changes and test results.

Keeping track of which changes you’ve tested will help you develop a set of best practices and will allow you to stay on track when trying out the list of changes you’re planning. When the time comes for you to redesign your site you’ll have a fairly good basis to start off with decent conversion rates.

The basic testing methods

Generally two testing models are used in web analytics: A/B testing and Multivariate testing. A/B testing involves changing one particular item on a page or a set of pages and measuring its effect on whatever you’re trying to improve. It’s like looking at before and after pictures, but instead of weight and carbs you’re looking at page views and unique visitors. OK, it’s a little less black and white than that, but that’s the basic idea. Multivariate testing logically consists of changing a number of items on the same page at a time.

Using the methods

A/B testing is generally the simplest way of trying out changes. It minimizes the factors that could have influenced an increase/decrease during testing, but because you’re limited to only one item this is a very time-consuming method. Multivariate testing allows you to do much more at a time but this method requires careful execution: The general rule is not to change items on the same page that could influence each other’s views/uses.

Test procedure

The trick here is to apply changes in periodical intervals or simultaneously, depending on your site’s traffic (and available time/resources). With a reasonable amount of traffic you could set testing schedules for weekly, sometimes even daily runs. It all depends on how much time you’re willing to allocate for this kind of ROI optimization. Allow sufficient time for tracking, analyzing and logging your changes to be able to efficiently and successfully plan your next step(s).