A / B Testing

You have an internet site and you sell perfume on the internet. Would you increase your rate of conversion if you use turquoise button instead of green button for the “Buy” icon on your website?

If you have a questions in your mind like this which you are curious about and want to try and see, you can find answers to these questions by doing A / B testing.

How is the A / B test done?

First of all, we need to establish the technical infrastructure of the A / B test on your website. This could be a technical infrastructure like Google Optimize, Google Analytics, Insider and VWO. After establishing the infrastructure, we need to determine the hypothesis of the A / B test. For example, your A / B hypothesis can be: “If I will add online chat module to my web site my sales will increase.” We need to test to confirm this hypothesis.

After the required technical installation, we create two versions with and without the chat module via the corresponding test tools. For example, 50% of the users who visit your site with the chat module and we provide service to 50% without the chat module. As a result of the test that lasts for 2 weeks, for example, the following table emerges.

Test ClicksPurchasingPurchasing
With Chat Module100023%2.3
Without Chat Module100017%1.7

According to these results, adding Chat module has a positive effect to increase your sales. So you can increase your sales by activating chat module now with peace of mind.

Things to pay attention when A / B test is carried out

  • Make sure that your traffic is distributed homogeneously while you are setting up your test. For example, make sure your Google / CPC traffic goes 50% to version A, and 50% goes to version B. Perform this check for all traffic sources such as Google Ads (AdWords), Facebook, Instagram, organic traffic, email traffic.
  • If there is no statistically significant difference, there may not be a winner occured in your A / B test. For example, if there are 23 purchase with 1000 clicks and 21 purchase with 1000 clicks, there is no statistically significant difference between them.
  • Do not test two simultaneous variables. For example, in version A, if the price is more affordable and there is red button and in version B, if the price is more expensive and there is blue button; Do not perform a test like this. If you need to test these two criteria simultaneously, create 4 variations instead of 2.
  • Decide the best solution to use when doing A / B testing.Once you understand the capabilities of the relevant A / B test platforms, make your decision.
  • Every subject you’re wondering may not be an A / B test. For example, a question such as “If I decrease the opening speed of my internet page from 9 to 3 seconds, does the conversion rate increase?” should not be the subject of the A / B test. Testing the areas again and again which were tested many time and which benefit the user will make you waste your time. Do not forget that you do not have to rediscover America.

If you think we can benefit you with our A / B testing skills, please contact us. Let’s decide together what we can do for you.