Web Analytics

What can/should be tested with multi variable testing?

When using a multi variable testing application like Google's new Website Optimizer there are a number of different page segments and/or pages you cant test.

First you should select what type of page you want to test. You could choose to test a campaign landing page, your home page, a lead generation form or a page in your check out process. When trying to determine where you should start looking for your low hanging fruit take a quick look at your web site's analytics.

You web analytics software (such as Google Analytics or Omniture's SiteCatalyst) will be able to point you in the right direction with respect to identifying pages that are causing your users the most difficulty.

Use your analytics software to determine which pages have the highest bounce rate (defined as pages that a user enters and exits from without visiting any other pages on your website), or highest exit rate, or pages that you feel are key to your business (these could be homepage, lead generation form or pages in your checkout process).

Once you have identified which page you want to test you then need to determine what it is that you are going to test. When you look at your page you need to view it as if it was created in segments. These segments are areas of the page that you would like to try alternative creatives or text in.

Try thinking of your page as consisting of the follow types of segments: Headlines, body content, images (both product and banner), calls to action, trust builders (privacy and return policies as well as logos like BBB online or Hacker Safe),…

You can also look at testing elements where the content or creative stays the same but you change the font size or color, bold, italics, underline,…

You can even test adding new elements to the page, not just changing elements that are already there. An example of this would be to add case studies or testimonials to your pages to try and increase the level of trust that you are creating with your users. Other new elements that an ecommerce site could test would be merchandising or cross selling. In this case you would like to find out if your new cross selling efforts helped increase you average order size.

Certain multi variable tests that you will want to run will require more data tracking than you will find from using Google's website optimizer alone. If you were to test the impact of making your telephone number larger you would need to be able to measure the impact this has on your phone orders. This is something that Web Optimizer on its own won't measure. If running a test like this you would also want to compare what happens to your online orders. If the larger telephone number leads to an increase in phone orders and an equivalent decrease in online orders then all you've likely done is increase your operating costs for the same sales volume.

If you are a lead generating site you would likely want to test your lead generation forms. How does the quantity and quality of your leads vary when you ask for more or less information?

So many questions and so many things to improve upon. Now with analytics and multi variable testing you have the ability to continually test and improve your online presence.

Coming soon from VKI Studios – Hippo Tango – the total website optimization service. Challenge the status quo and strive for continuous improvement of you web property with the help of analytics, multi variable testing, web usability, design/branding and integrated online marketing campaigns.

Cardinal Path

Share
Published by
Cardinal Path

Recent Posts

Optimizing user experiences with Digital Experience Analytics (DXA) platforms

As consumers become increasingly digitally savvy, and more and more brand touchpoints take place online,…

2 months ago

Enabling Value-Based Bidding with Google Tightlock

Marketers are on a constant journey to optimize the efficiency of paid search advertising. In…

3 months ago

Resolving “Unassigned” Traffic in GA4

Unassigned traffic in Google Analytics 4 (GA4) can be frustrating for data analysts to deal…

3 months ago

This website uses cookies.