11 February 2022

How do you get the best result from your A/B test?

️ Blog
14 min. reading time

Do you want to know what an A/B test is and how you can increase website traffic with it? Then read on. We explain to you what an A/B test is exactly, how you should and should not apply it and how you can improve your website with it. These tests are aimed at testing two options. The goal is to find out which option works best for your target audience. You get to know your target group better, so that you can give your visitors the right information sooner and achieve higher conversion figures. By testing you ensure that your adjustments are scientifically sound and not purely intuitive.

What is an A/B test?

Back to the beginning: what exactly is an A/B test? Simply put, an A/B test is a way to compare two options, such as two advertisements, two newsletters or two websites. In the rest of this blog we will focus on A/B testing for websites. You create two variants of the site for this: some of the visitors receive variant A, the rest variant B. You can make the variation in the call-to-actions, push notifications, location of elements or colors used. Be smart and make sure that the variants only differ in one point, for example a website with a pop-up offering a whitepaper and a website with a call-to-action at the end to download the same whitepaper. You then compare the conversions of both variants and then you know which variant works better.

If the variants differ on several points, you do not know what causes the difference in conversion. Therefore, perform several A/B tests if you want to have information about several differences. The difference can be relatively large, but also very small. For example, consider testing two different color schemes for the entire website or testing two colors of a specific button: do people press a light blue or a dark blue button faster?
A / B test

What does that look like in concrete terms?

Better stolen well, than poorly conceived: our friends from Hubspot give on their website some interesting examples of A/B testing† For example, they have experimented with the search bar on their own website. People who use the search bar have higher conversion rates than people who don't. That's why Hubspot decided to test its visual presence. They tested different variants for this, which all turned out to increase the conversion. The variant in which the search bar displays the text 'search the blog' instead of, for example, 'search by topic' worked best.

A / B test
A / B test

Interesting material is landing page banners. These are of course very important because they are the first introduction to the website. Hubspot gives the example of Humana, an American health insurance company. By shortening and clarifying the text on the banner, they were able to increase their CTR by 433%. The use of the word 'shop' in the CTA also resulted in a 192% increase in CTR. A/B testing can therefore make a big difference. That is why it is essential to know how to use an A/B test really well.


A / B test

Smart testing

Many companies perform tests because they think it should, without really thinking it through. As a result, they do not get the most out of it. Make sure you test things that are really useful to know. Don't test for the sake of testing: you're wasting your time. You can imagine that a candy pink background that makes your eyes pop out of the sockets will not have a positive effect on the number of visitors that stay longer on your website. But don't make any changes to the website without knowing what the result will be, just because your manager thinks it's a good idea. So test adjustments that you suspect have a positive effect to make sure your hypothesis is correct. This gives you the tools to make the adjustments on your website data-driven, instead of intuitive.

Coincidence or statistic?

Now about the numbers. To publish the two versions and get them evenly distributed to your audience, use a plugin like Google Optimize. If you know which variants you want to test, think about which conversions you should measure for the test, such as clicks, downloads or orders. Think about what your target is and which metrics are important for this. Once you have this set up, you can launch the two versions and keep track of their results. It is important that these results are statistically significant. Being able to write one extra quote with variant B does not mean that variant B is better. It just means you got lucky. The P-value of your test results may therefore be a maximum of 5%. The standard deviation is as small as possible. Based on that, determine how many visitors you need to make your test successful. Based on the average number of visitors, you can then determine the duration of your test. It is advisable to go for tests that you can complete in a short time, because then you can optimize your website the faster. With a product page with about 5.000 visitors per week, you can already have a significant result within a week.

What do you do with the results?

After the test, you will know which variant works better and therefore yields more conversions. You will of course use this variant. If there is a greater chance that people will click through when using banner 2 instead of banner 1, then you will of course use banner 2. In the same way you choose the search bar with 'search the blog' as text if this appears to work better than 'search by topic'. The results therefore give you the opportunity to create the best version or the best working version of your website.

What can Social Brothers do for you?

If you want to stay informed of the latest developments, be sure to subscribe to our newsletter. If you don't want to get started yourself, don't hesitate Contact up. We are happy to investigate what the most optimal version of your website is.

Steven Founder[email protected]06-20413957
I practiceWhatsapp

Inspired? The following topics have been offered

Online Marketing
Show all inspiration