Conversion Guru - Online Secrets to Success

Monday, March 20, 2006

A/B Testing - Google Style
(Google Adwords and Google Analytics accounts required)

Hopefully you already know about the wonders of split testing and why it is so important, and you are already testing on your site. If you are part of a large company and have access to a larger IT infrastructure, hopefully you are doing some more complex A/B testing. If you are really ahead of your competitors, you may be performing some multivariable tests.

However, if you are a small company or haven't had the time to implement the one thing that could probably have the greatest impact on the long-term success of your company - split testing, you can do it EXTREMELY easily - all within Google.

Fortunately, Google has a desire to help people find relevant results. They not only try to help produce the most relevant results with the help of their search algorithm, but they actually give web site owners the ability to help generate pages and content that most interest the user. They give people the ability to make the actual pages more relevant for the user.

Even though split testing is very easy to implement, Google makes it even easier for anyone to do it - regardless of technical abilities. There are a variety of ways to do split testing, but this article
will focus on using Google tools exclusively. However, this option only works for traffic from Adwords. The next article will talk about other options available to you. Here is a brief tutorial:

  1. Look at some of the most important elements on your landing page (Header, Call To Action Background color, etc.) and decide what you would like to test.
  2. Make two pages. The first page will be your existing page and the second will be the same page with one element changed. Change only one element at a time. It is possible to do multivariable testing where you can test multiple elements on a single test, but that is for another article.

  3. Log into your Adwords account and create a second ad within your AdGroup. The new ad should be exactly the same as your existing ad. However, the second ad will link to your test page, while the first ad still points to your original page.

  4. When entering the URL, make sure you are using the Google Analytics tracking URL. Create the URL's with the URL Builder Tool and make sure you include the utm_content variable that will differentiate the two landing pages. The two URL's should be the same with two exceptions: 1) The landing page will be different, and 2) the utm_content will be different).

  5. In Adwords, go to your Campaign Settings and select the campaign that contains your test AdGroup. Click on "Edit Settings" and make sure "Ad rotation: Show better ads more often" is unchecked. If not, it may not display the ads an equal number of times. It is vital to make sure your two tests are identical, with the exception of the one changed element.

Your test is now set up. Once the ads have run for a while, you can view the results in your Google Analytics account. To view the data, go under the "Content Optimization" folder and select "Ad Version Testing,” then "A/B Testing.” It should display the data from your tests. The variables you entered for utm_content will be displayed, and you will be able to see which version is performing better, assuming you have goals set up. You can also view other data to see how they are performing. Since you will have different landing pages, you can view other reports to see how the pages are performing in areas other than your goals. For example, bounce rates and navigation after getting to the pages are important as well.

One thing you will need to keep in mind is that you need to let the test run until you have a large enough sample size. It is generally recommended to wait until you have at least 50 actions (form being filled out, a sale, etc) for each page or at least 1,000 visitors. However, it also depends on how close the two pages are performing compared to each other. If both pages have received 1,000 views and version A has recorded 50 actions (leads, sales, etc) and version B has 25 actions (leads/sales, etc), this is plenty of data to let you know that version A is better. It is not necessarily important to know how much better one version performs compared to the other. Your goal is to simply continue testing new versions and continually get better. Small improvements do not take long to add up to large improvements.

There are a variety of drawbacks with using this method for split testing. First, you are not able to test different ads to attempt to improve Click Through Rates. Second, setting up a new test requires you to log into Adwords and change the landing page URL anytime you want to do a new test. Also, the test is limited to traffic you are receiving from Adwords, which may be a very small percentage of your overall traffic and will require you to test much longer to get a large enough sample size.

Split testing is a rapidly growing technique marketers are using to improve their websites for visitors. By taking the short amount of time required to do some simple A/B testing on your landing page, you can increase your conversions and profits drastically.


Post a Comment

<< Home