How to Conduct an A/B Testing in Design

6minutes read
how to do ab testing

Assumptions built on guesses aren’t always aimed at the result. It’s necessary to take definite action to achieve success. If we’re talking about web design, success is when your audience actively uses your product. Thus, the object of research is our target audience, and we constantly try to improve our product for them. So, how to get maximum results without wasting time and get the answer to the most exercising issues: how do I increase the conversion rate of my website?

In this article, we’ll discuss the A/B testing method in web design and highlight the essential tips for conducting this test to get visible results.

What is A/B testing?

A/B test experiment design
a/b testing in web design

A/B testing is a research method that shows us which of the two proposed options would be more advantageous. The primary purpose of such testing is to make decisions based on facts, not surmise.

A/B testing is used when you want to implement a new feature and need to test its effectiveness. For example, we have a site with good traffic, but the designer has decided to add a new widget to subscribe to news. We intuitively assume that everyone will actively use the new feature, then, expect an increase in traffic and readers. But our assumptions can be distinguished from the views of users. So, we run an A/B test to confirm or refute our expectations.

The essence of A/B testing

how to design A/B test
a/b test example

The essence of the method is simple. First, you create page A and its copy — page B. Then, on page B, you make changes to the elements you want to test. This could be a header or a micro-interaction button. Then you show page A to one half of the users and page B to the other and observe which page has reached its intended goal and where the conversion rate is higher.

Read more about conversion-centered design in our blog.

You can test everything on the website, but there are several elements that will have the most noticeable test results.

Elements that can be A/B tested:

  • Page layout options;
  • Titles: content, location on the page, design;
  • Subheadings;
  • Texts;
  • Call to action text;
  • Images, its size;
  • Links and emails.
A/B test elements
a/b testing elements in web design

Types of A/B testing

There are three essential types to how to run an A/B test:

1) Split URL testing

how to run an A/B test
a/b test types

This is testing two drastically different versions of a website with different URLs. Your website traffic is evenly split between the two versions: the original page URL and the new page URL, which helps you figure out the most profitable one with high conversion.

Also, you can check the website design services we provided.

2) Multivariate testing (MVT)

how to set up an A/B test
schematic of multivariate a/b test

This type is more complicated than the A/B test. During the test, explore several variable pages to calculate which combination works better than the others. With multivariate testing, you can solve the problem much faster and don’t run many simple A/B tests.

3) Multipage testing

how to perform A/B testing
multipage test type

This is a form of testing to determine changes on numerous pages at once. For example, you can create copies of all your funnel pages and test them simultaneously in multipage testing.

Tips on How to Conduct an A/B Testing

If you make a mistake when running an A/B test, you can get worse or unpredictable results. We gather TOP ten tips on how to do A/B testing. Let’s analyze.

how to conduct an A/B testing
a/b testing in web design

Determine your goals

Before you start working on the A/B test design, set clear goals. These can be things you want to check or improve by the end of the test.

Example of goal: you need to increase downloads of your product through the website. So, using the A/B test, you’ll check whether changing the size, color, or text of the “Download” button will cause more downloads by the users?

Define the hypothesis

The hypothesis is necessary to your business, so you don’t work blindly. Each process has an eventual result, and it either meets your expectations or not. For example, build a hypothesis to do an A/B test with a specific task.

An example of a hypothesis: “I think that by changing the text of the call-to-action button “Download” to “Download Free” and changing its color from white to blue, it’ll raise the number of downloads.

Research your website traffic

Research your website traffic before running the A/B test. Its criteria are responsible for the effectiveness and accuracy of the result.

If your website traffic is relatively high, 5000 or more hits, you can run A/B tests often and quickly and focus on minimal changes that affect vital metrics.

If your website’s traffic is small, from 300 hits during the day, conducting an A/B test is necessary to make significant changes to calculate what design will be beneficial and increase traffic.

Decide what website element you are testing

The item that you’re testing is the item on which you build your hypothesis. Important rule: you can try one item in one test. If you include several research items in one test, you can’t understand which of them gives the result.

Calculate the sample

when to do A/B testing
calculate the sample

It would help count how many people can visit your website to get statistically significant A/B testing results. The sample number will determine what changes you want to see. The more global it changes, the fewer users you need. Online sample calculators: Optimizely, AB Tasty, Unbounce, AB Test Guide.

Set A/B testing deadlines

The minimum duration of the A/B test is one week. This is because users behave differently depending on the day of the week, and hasty results can be incorrect. Therefore, the optimal duration of the research is 10-14 days. Especially if your A/B test aims to check whether people make a purchase or not by changing certain website design elements. Usually, people require time to think about a purchase, so 10-14 days is a normal time to see if your hypothesis works or not.

Consider the contributing factors

Associated factors can be different situations that change the usual course of events. These can include winter holidays, summer vacations, or even changes in currency exchange rates. All of these factors can affect the test results.

Make sure that the conditions are equal for all test versions

As we remember, in testing, the target audience is divided in half between the two test subjects. Therefore, these two halves of the user must be as similar as possible so that only your experiment will affect the test result.

Observe the progress of the test

After you run an A/B test, keep an eye on the results. If you see a big difference in conversion between the two samples during the experiment, ensure that all features, buttons, and links are working.

Get feedback from users

The A/B test is quantitative research. It’ll show you the numbers, but it won’t explain why users prefer this option over the other. So instead, use the qualitative method of gathering information — survey to keep in touch with your users and understand their behavior. It’ll dramatically improve the test results.


A/B testing is an effective tool that helps to get visible results. It’ll understand what users like and what changes need to be resorted to achieve the desired result. Therefore, it’s essential to conduct the test with a hypothesis that you want to test.

Business is always about users. And the A/B test helps to keep in contact with them. So if you’ve decided to work on improving your website, you’re on the right path. Contact us, and we’ll help you to solve a problem. Check out the finished projects on Behance and Dribbble. And also take a look at the services we provide.

Rate this article

20 ratings
Average: 4.9 out of 5

If you like what we write, we recommend subscribing to our mailing list to always be aware of new publications.

Do you have any questions? We tried to answer most of them!