A/B testing: what has changed in 2017

The first quarter of the year is over, but this does not mean that you will no longer be able to improve the performance of your company in 2017. After all, there are still 9 months ahead. So, include in your plan for the current year conducting A / B tests that will help optimize conversion and attract new users (or retain old ones). And we will give you six useful tips on how to do it right.

Tip #1. Stop doing A/A tests

This type of test is done on the same page. The method is good for controlling experimental settings, but nothing more. By conducting such tests, you spend exactly the same amount of time that you could devote to experiments to improve usability and increase conversions. In addition, you risk losing some of your customers if the experiment fails.

The same tasks that A / A testing solves are solved by constant monitoring, user surveys, and analysis of certain statistical information. Moreover, they can be more effective than "single-page" A/A testing.

Council number 2. Tests without results are also needed

Some tests can be called inconclusive and even failed. But they are also needed. Yes, they require resources (including temporary ones), but before testing begins, you cannot know what results it will lead to. In addition, any testing leaves a lot of data for analysis at the end.

You can use the information received to better understand your target audience, evaluate traffic sources, the type of devices from which users access your page, and find out certain demographic and geographic information. In addition, you may find that one type of page is more effective for regular customers, while the other is more effective for those who see your offer for the first time.

Remember that testing is a lengthy process. If your hypothesis doesn't work right away, try to refine it before declaring it invalid.

And always pay attention to micro conversions, the small steps your users take on their way to an order or a subscription. If you don’t want to set global goals for testing, you can always see how changes affect micro conversions, and then improve the result, increase it.

In addition, you should always remember that the purpose of an A / B test is not only to identify the result that works best, but also to get to know the target audience as closely as possible. All the data obtained during the study can be used to develop the marketing concept of the company.

Council number 3. Repeated A/B tests are required

Do not rely solely on static significance. Consider the average purchase cycle when choosing the duration of the test. The minimum duration of testing is one full cycle. Then you need to wait a while (from a week to a quarter) and test again. This will help improve the statistical sample and get a more complete result. After all, a variety of factors influence user behavior, from the time of year to the marketing policy of your company or competitors.

Tip number 4. Engage Feedback

What you can get by analyzing the test results is good. But it is better if all statistical information is supplemented by real feedback from users.

Feedback is needed, it helps to find out why customers choose this option. So build your work like this:

• conduct an A/B test to determine which pages have the highest conversion rate;
• together with testing, make a survey that helps to understand why users choose certain pages;
• make the necessary changes on the site after the "winner" is approved;
• conduct the survey again to ask users how they understand and like the changes.

In this case, the results of the experiment will be as reliable as possible and will have the highest value.

Council number 5. Consider the interests of mobile users

You will get the best result from testing if you separate mobile and desktop traffic. The thing is that the user of mobile devices cannot concentrate for a long time - most likely, he is away from home, on the road. If you do not segment traffic, you can make changes that are unacceptable for mobile device users.

But tablets and smartphones are the most popular gadgets for finding information. And people are more likely to buy from desktop devices.

Council number 6. Regularity is important

If you don't test, you miss out on effective improvements. It is better to get started right now and experiment with constant testing, rather than miss this opportunity altogether. Improvements based on the results of tests and surveys will bring you an increase in conversions!