How A/B testing can help you market better

September 9, 2016 - 4 minutes read

Everyone uses A/B testing in their lives to make better decisions. Consider the route you take to office. You might have initially tried out different routes to get there considering factors of distance, time, traffic, road condition etc. By analysing different variables, you have inadvertently used A/B testing to improve your life.  Marketing teams too, put in a lot of time and money doing A/B testing to perfect marketing decisions. A/B testing is a form of CRO (Customer Response Optimisation) that tests one variable in a marketing content against another.


How does A/B testing work?

An A/B test involves two similar versions of one piece of content with only one variable changed. There are the two variants A and B i.e. the control and the variant which will be presented to two similarly sized audiences. Responses from these groups help us judge which variation of the content performed better.

Imagine you have a sign-up form on your landing page and you’re not sure which background image will attract a higher sign-up rate. You’d use A/B testing to present different versions of the landing page to two different audiences. This is exactly what Bernie Sanders did as part of his 2016 presidential campaign.


Source: medium.combernie2

Comparing responses on both webpages helped them optimise the webpage to attract a larger audience.

Pick the variables you want to test

Anything in your marketing material is up for testing. As you test more variables within your content, the metrics you receive from each test will help you tailor better content and market it to a wider audience. The likely variables for testing include:

  • Landing page images
  • E-mail copy
  • Call-to-action button
  • Text messages
  • Promotional offers

Don’t hurry the results

A/B tests should be given enough time to run ranging from a few days to a couple of weeks. Decide on a suitable time range depends on the amount of traffic that your website usually receives. Giving your A/B test enough time to run ensures that your results come from a larger pool of users making them more statistically accurate. Be careful to not let the test run for too long. This can bring in other unknown variables into the equation making the process more complicated. If you are in doubt about the accuracy of test results it is perfectly okay to run a retest.

Run one test at a time

Running multiple A/B tests at a time can lead to confusion about the results. Imagine A/B testing your landing page background while running a separate test on your sign-up button. When the leads start coming in you’ll have some trouble figuring out which test brought you the leads.

But in cases where you are testing multiple variants of similar content, maybe three different variations of a sign-up button, its okay to run an A/B test. Splitting your visitors into three groups rather than two makes sense in this scenario.

While A/B testing commonly uses two variables for comparison, certain situations require you to run tests with multiple variables. This is known as multivariate testing. We’ll be sharing more information about multivariate testing in upcoming blogs.

Tags: , ,