How to AB Test Like a Star

By Anthony Coe | Google AdWords

May 05

AB testing is powerful landing page optimisation methodology. Any advertiser looking to increase the effectiveness of their landing pages should be testing constantly. In this post, we explain what AB testing is, its importance for Google Adwords as well as the simple steps required to set up your first experiment.

What is AB Testing?

AB Testing is used extensively to optimise online advertising, email and inbound marketing campaigns.

It is landing page optimisation technique used to improve the % of visitors that convert to leads or customers.

AB testing involves running simultaneous experiments using two identical landing pages with 1 variant, which can be any element on the page, such as a headline, image, string of text, video, call to action, etc.

AB testing

The experiment entails sending an equal amount of traffic to each page and then reviewing conversion & ROI statistics to see which version performs best. Below is an example:

AB Testing Table 1

The version that performs the best will then become Version A for the next experiment. This is sometimes referred to as the control version or champion page.

Constructing your Hypothesis

The first step is to formulate your hypothesis, an assumption made to draw out conclusions via testing.

Before deciding what to test, you need to define your conversion goal for a given page and make a list of the supposed problems in achieving this goal. To aid you in creating this problem list, consider getting user feedback from friends, colleagues or via remote user testing (i.e.

Once you have your problem list. Create a list of proposed solutions or hypotheses in the adjacent column:

ab testing problem hypothesis

Order your list starting with the hypothesis with the highest anticipated results and work backwards, testing in one in turn. If your first experiment is successful, Version B becomes Version A for your next experiment and so on.

Conducting your experiment

  1. Firstly, ensure you have analytics goals for your landing page.
  2. Start off by creating Version B. Remember that other than the one variant (i.e. the button colour); Version B will be identical to Version A.
  3. Send equal proportions of traffic to each version. Google Adwords is probably one of the better ways to achieve this, as it allows you control of the budgets, the quality of visitors, keywords and ads. With Google Adwords, setup two ads with the same copy going to different landing pages.
  4. There is no best practice duration for running your experiment, but we advise no less than a week.

Recording your Data

Once you have conducted completed experiment, record your data in a spreadsheet:

ab testing table 3

In this post, we have considered the importance of AB testing for advertisers as well as the process of setting up your first experiment, including constructing your hypothesis, conducting your experiment and recording data.

By Anthony Coe