Semalt: How To Conduct A/B Testing On A Website?

If you want to improve website conversion or get more leads for a given activity, you should be familiar with the concept of A / B testing. After all, it is nothing more than a comparative research method used to maximize revenues or achieve goals for a specific campaign.

What are A / B tests?

Have you ever wondered how the decision-making process usually works for a specific campaign, project or investment in your company? Most projects do not have a designated status of activities and a developed implementation scheme - at least that's what managers say when asked for their opinion. Usually, the main assumptions are created by a project manager who acts quite instinctively, sometimes also guided by experience, but it may be different and not always tailored to the needs of a given moment and a specific client.

Considering the possibility of mistakes, such a scheme does not give any certainty of success, does not allow you to quickly eliminate errors, and does not guarantee that a specific, not always consistent way of thinking will be the same as the preferences of customers. This is why it is so important to implement A / B tests, which not only give a clear picture of the situation, but also quickly disqualify the danger of a bad investment and wrongly chosen direction of distribution or promotion in a given budget.

In order to understand well what A / B tests are, it is worth providing at least a few important examples where their implementation can bring real benefits. So what can you test using this method?

Both administrators and website owners themselves can face unsatisfactory results many times. We are talking about conversion, profits or traffic generation. In such a situation, will the use of a new version of a tool be appropriate? Of course! Thanks to this, we can, not only attract a potential customer, but also show that, as a company, we are developing and not lagging behind the competition. And for that, you can use the DSD Tool, to generate traffic and conversion.

As you know, aesthetic issues are quite important because they usually keep a potential customer from our offer for longer. What to do to make the user actually stay with us for longer? A / B testing is quite common today and consists mainly in preparing two variants of a website that differ in the selected element - the one that turns out to be more favourable is left on the website. Thanks to these observations, we can learn about the behaviour of users, which affects the further understanding of the customer's path on our website. It is worth remembering about the analysis of statistics from the Dedicated SEO Dashboard or Google Analytics, which will be key in performing A / B tests on the website. Sometimes small changes bring spectacular results!

Banner appearance

Graphical changes for a specific message may indicate which elements are definitely more attracting the eyes and attention of users, and which are on the contrary. Sometimes, it is enough to move the button - call to action button here - to another place or use the eye-catching face of the campaign to get completely different numbers from a click-byte.

But that's not all, because as a graphical approach to the banner topic in A / B tests, the meaning and eloquence of messages should also be considered. Different font, bigger typeface or banner text colour - each element can have a greater or lesser impact on the clicks.

Therefore, use various forms of banners and distribute them simultaneously to specific target groups. On this basis, you will draw conclusions on how to construct subsequent messages and, at the same time, reduce the likelihood of developing materials not fully profiled with your activity and user preferences. Test each time you launch a click-byte campaign. Over time, you will come to the appropriate conclusions.

Newsletter subscription

When you release an e-book, you usually have high hopes for the success of your project. Then it turns out that the effects of your work are different, because… there are few users who subscribe to the newsletter and download the e-book. In order to test different variants for the record, it is worth using A / B tests, which in this case can give huge power in obtaining more leads.

Sometimes it is enough to change the main message to convert several times more subscriptions to the newsletter. Test your word power by taking into account numbers, adjectives, special characters and more. Take a look at these variants:
  • How to get 50% more customers in a month?
  • How to get 30,000 more clients in a month?
  • How to get half more customers in a month?
  • Can you earn more?
  • Can you earn more?
  • How to earn more?
  • See how to earn more!
These variants, although apparently very similar, differ. You will be surprised, but such small changes can produce different results for your campaign.


In A / B tests, it is also possible to verify not only whether the user is navigating correctly through the stages of the order, but most of all, whether the entire path is properly designed. Sometimes eliminating one of the burdensome stages of procurement can significantly affect the generated revenues.

Did you know customers are reluctant to provide businesses with a phone number? Leave the field with such data only when it is really necessary.

A / B test plan

In order to obtain correct A / B test data, the activities should always be carried out at the same time, parallel and by target groups. For example, you can divide your users into two groups and send them two prepared messages: A and B. You can also divide the clients into three groups and first present the message to two groups, and then send a better converting version to the third group. Then you have a good chance to verify the revenues from the selected version of the campaign or website settings. You can also send a message to the entire group of users and compare the test results with the old period.

In general, the A / B testing principle is quite simple. You check which option brings better results, e.g. more subscriptions to the newsletter, higher CTR or better income. You can immediately verify how the applied change works - this is a huge advantage of this method regarding introducing updates to the website.

Remember to take into account seasonality, changes in the offer, temporary technical problems or activities carried out by competitors when verifying the results derived from A / B tests. Also, avoid conducting tests at breakthrough moments - when there are sales campaigns or during the holidays. Then the results may not be reliable.

Remember! In your action plan, include time to:
  • conduct an analysis (with a tool like the Dedicated SEO Dashboard) - to know what exactly to test; do not waste time on tests that are not of great importance and will not change much in the context of sales in the long run;
  • develop a hypothesis - what exactly do you want to check, what goal you assume and what solutions may bring benefits;
  • configure an appropriate A / B testing tool;
  • test the performance for individual versions;
  • monitor the test to make sure it works as planned;
  • summarize conclusions and possible test optimization.
A / B testing is hardly recommended for small sites. Here, your main action should be primarily to increase traffic, because the larger the scale of the business, the more profitable it is to conduct tests. What's more, the more traffic the website receives, the shorter the testing time. You can check more solutions and have a better chance of increasing conversions. Even a small increase in your conversion rate can mean hundreds of thousands of dollars of additional income.

A / B test: A common mistake that has a key impact on the result

A common mistake that has a key impact on A / B test results is completing them too early or entering them at a time when the site does not experience high traffic. The calculator created by Evan Miller can be used to calculate the length of time needed to perform the A / B test. From it, you will find out how many users per version of the site you need to get reliable results.

What tools will help to conduct A / B testing on the website?

Choosing a test tool for A / B testing is of utmost importance for obtaining correct results. In your optimization plans, you can include a tool like Google Optimize - a free A / B testing tool - perfect if you're just getting started with this type of analysis.

A / B test tool options are usually very similar and are based on:
  • redirecting to a different URL when you test landing pages located at different addresses;
  • HTML element replacement on the browser side.


Do the A / B tests work? It is enough to look at the examples to unequivocally say that they do. There are also interesting case studies on the website. Perhaps you will discover something there that will perfectly harmonize with your business. Good luck!

If you need even more information on optimization, please contact us.