A/B testing

To get the best results from your ads, try variations in proposition, target keywords, copy and title.  The ‘Create a Similar Ad’ feature by Facebook is designed to speed up the process whereas AdWords helps in optimising their display and measure the relative success of adverts in an ad group. This technique is useful for more than just ads, though. App features, website marketing, messages, call to action and email campaigns can be tested against to figure to their effectiveness.

Visual Website Optimizer and Google Website Optimizer are among those apps designed to ease the A/B testing process.

There are four important points to remember when you create and run your tests:

  1. You need to understand the purpose as well as the effect to keep the variation in each test to a single element, such as the image or title.
  2. In most cases you will find that, while conducting the A/B tests for sign up buttons, adverts and call to action is done to measure the click-through rate as the default goal. Though more clicks never imply higher revenue. One should rather focus on variations that lead to more profit.
  3. Numbers can be really confusing. The solution lies in the confidence or statistical significance which is part of A/B test tool for preferably accurate results.
  4. See, the statistical confidence may keep varying. Hence, fix the number of tests up-front. Usually you will find certain sample duration or size in most of the tools. You can also use the Google test as a tool.

Other than variation in words, one can also check positions, sizes, images, colours and prices. Since all these will help you to better understand the attributes which can make valid test subjects and influence the customer conversion. There is no denying that A/B tests are powerful, but let’s not base every design decision on them. They will measure the change brought in by one variable and its impact and avoid the subtle changes.


Leave a Reply