All Things Digital

12 A/B Split Testing Mistakes Businesses Make All The Time

Whether you’re updating your web presence or wondering whether it might be time to do an update, A/B split testing can provide valuable insight into how your customers really feel. When done properly, A/B split testing drives critical business decisions. Today, we’re going to look at 12 A/B Split Testing Mistakes Businesses Make All The Time and how to avoid making those mistakes.

  1. Skip A/B split testing altogether. Maybe this goes without saying, but maybe not. It may be tempting to go with your gut on this one, even though you know your gut may be wrong.
  2. Test for too many variables. Ideally, A/B split testing should test for only a single variable, a single difference in your electronic text. Even if you need to test for multiple variations at one time, consider phasing those changes over several tests. If you change too many things, for example the color of the Submit button and the content of your Call to Action and the introductory paragraph, you won’t know which part or parts work, together or separately. Your tests should include multiple variations in such a case, but never all at once.
  3. Test for too few variables. On the other side of the coin, you may want to say, well, you are pretty sure that most of the changes will be okay, but you’re just unsure about this one. Here again, the A/B split testing will give you false data. A change that looks good to customers in one context may fail in another one, or vice versa.
  4. Test over too short a period of time. It may be tempting to get those tests done quickly, but don’t. Testing over too short a period will you give you bad results. You will not account for the breadth of your client base.
  5. Fail to comprehensively test navigation buttons and links for each option. If any of the links are incorrect, you will get bad data (due to never knowing for sure how many people clicked). You will also get some frustrated customers.
  6. Fail to comprehensively read the text of altered messages. Make sure that you have completely read over any text which has been edited. Typos happen, especially where the text is nearly identical with only minor changes.
  7. Fail to test layout changes on multiple browsers and systems. Before releasing an A/B split test into the ether, you need to use the same due diligence with any change. Make sure text is not now running over images, that images render properly and attractively on multiple browsers, and so on.
  8. Not conducting A/B split testing based on varied demographics. Track relevant user data such as gender, geographic location, and age, and then parse your A/B split testing in order to get data based on these different groups. That way, you can vary user experiences depending on what demographic you’re targeting.
  9. Not revisiting A/B split testing periodically. A/B split testing is not a one-and-done operation, because in modern web development there is no such thing. Test once, use what works, and then test again later to see if anything has changed.
  10. Not tracking A/B split testing throughout the user experience. A/B split tests should follow customers throughout the remainder of their visit. Click-through rate matters, but so does customer conversion and customer satisfaction. Did they buy anything? How much did they spend? Were there any complaints – or compliments?
  11. Not analyzing A/B split testing results correctly. Ideally, you should conduct A/B split testing over a broad range of users, accounting for relevant demographics and tracking how they respond to changes throughout their visit. Once this is done, you will have results, but those results can be difficult to quantify. Make sure you are properly quantifying all relevant data.
  12. Over-reliance on A/B split testing. A/B split testing alone is not enough, without bringing your own creativity to the table. After all, if you don’t have brilliant ideas to test, the best A/B testing in the world won’t help you all that much. As Steve Jobs put it, “It’s not the consumers’ job to know what they want.” That, of course, is your job.

Are you considering A/B split testing? Do you have questions about how to improve your inbound marketing campaigns? Contact us today and see what Highpurple can do for you!