A\/B testing is your ticket to understanding what works and what doesn\u2019t. It allows you to get into your users\u2019 heads in a quantifiable way. As a UX designer, there\u2019s nothing more valuable than that.<\/p>\n
A\/B testing is a standard method used to understand user preferences quantifiably. Designers can also use it to solve a debate within the design team. When it comes to user experience, relying on data and not intuition is an important part of the design process<\/a>.<\/p>\n Testing in the design phase and during the rollout of new features helps you create the best possible product.\u00a0 In doing so, you may realise that a feature is unnecessary or not beneficial after testing it. Or, you could find that you\u2019ve completely missed a feature your users really need.<\/p>\n In this guide, you\u2019ll find out what A\/B testing is, how to conduct one and why you should A\/B test.<\/p>\n [GET CERTIFIED IN UX]<\/p>\n Take our Professional Diploma in UX Design course<\/span>\n <\/a>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n A\/B testing compares two versions of a product, web page, app or concept against each other to determine which performs better. It\u2019s a quantitative way of finding the best possible version of your work.<\/p>\n Also known as split testing or bucket testing<\/a>, A\/B testing is basically an experiment where two options are shown to users at random. Then, marketers, developers, designers or analysts use statistical analysis to determine which variation works better for their outcome goals. A\/B testing is all about finding out what the user prefers.<\/p>\n A\/B testing removes the guesswork from user experience (UX), user interface (UI) design, as well as many other professions like marketing and data analytics. A\/B testing directly measures the impact of any changes you make so that you can ensure you\u2019re actually creating an optimal result.<\/p>\n Testing your designs<\/a> helps you improve your product and ensure it\u2019s what people want. In the long run, A\/B testing also saves companies money \u2013 especially if they conduct it in the design phase. Testing often prevents mistakes or subpar designs before they\u2019ve invested in development.<\/p>\n A\/B testing isn\u2019t just for the initial design phase of a project. Even after a product has been launched and developed, you can still conduct A\/B testing. It helps designers find out what users want, optimise and iterate.<\/p>\n Here are some reasons you might want to conduct an A\/B test:<\/p>\n It\u2019s essential to conduct targeted A\/B testing. That means changing only one aspect of your design per test. This will be the \u201cvariable\u201d. You\u2019ll have to determine what you actually want to test.<\/p>\n For example, you might change the size of one button, the colour of one button, the location of one submission box, the CTA copy on one button… You get the idea \u2013 test only one variable.<\/p>\n You may want to consider testing things that will have a big impact on user experience or that you don\u2019t have enough data to understand. Look for high bounce rates, low conversion rates, high traffic areas, abandon points or common error points.<\/p>\n Here are some variables you might want to test:<\/p>\n – Layout – Content – Fonts, which includes:<\/p>\n – Colours, which includes:<\/p>\n Images, including:<\/p>\n [GET CERTIFIED IN USER RESEARCH]<\/p>\n Take our Professional Certificate in User Research<\/span>\n <\/a>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n To determine which version of a design is better, two versions are created at the same time. Then, version A will be shown to one half of the target audience, while the other version will be shown to the other half of the audience. The option that prompts users to take the desired action is the winner of the test and usually the design that gets implemented.<\/p>\n Let\u2019s learn how to conduct an A\/B test step-by-step with examples.<\/p>\n Before conducting your test, you need to know why you\u2019re running the test. Set a goal that you want to achieve. That way, you can determine which version to continue using or testing. Be sure to gather any data you can before your test so that you can quantifiably see changes.<\/p>\n Examples:<\/p>\n Think about the goal you want to achieve. Consider what single aspect of your design you could change to get closer to that goal. For a list of ideas, scroll back up to \u201cWhat to A\/B Test.\u201d<\/p>\n Examples:<\/p>\n Make an assumption on which version do you think will perform better. This will help you solidify why you\u2019re conducting the test and critically consider your designs. What could be preventing the result you\u2019re trying to accomplish? Try writing out your hypothesis as a single sentence.<\/p>\n Examples:<\/p>\n Create two versions of your design to test: A and B. This could be a prototype of almost any fidelity. Then, determine your sample audience size to get statistically significant responses. Then, randomly expose half of the target audience to version A and the other half of the target audience to version B.<\/p>\n Monitor the test to make sure it\u2019s working properly but don\u2019t look at the results. You\u2019ll need to wait for the test to be complete before looking at the results. It\u2019s tempting to watch the results and assume a winner early on but you need to give it time.<\/p>\n The test must run long enough to produce a statistically significant, meaningful result. Wait until you\u2019ve reached the minimum sample size you determined in the last step before looking at the data. Meaningful results can take anywhere from 48 hours up to two months, depending on your chosen sample size.<\/p>\n Once the results are in, crown the clear winner by implementing the results.<\/p>\n Examples:<\/p>\n When you\u2019re analysing your test results, you first need to make sure they\u2019re statistically significant. Statistical significance is a threshold that determines the level of certainty that the results of a test are not due to a sampling error. It accounts for the uncertainty that comes along with randomised tests.<\/p>\n You\u2019ll rely on the number of users and the number of conversions for each of your variations to determine statistical significance. The goal is to reach 90% or more statistical significance before assuming meaning from an A\/B test.<\/p>\n Once you determine whether your results are statistically significant, you can decide how you\u2019ll use your data. Maybe version B performed better and you implement it. Maybe your original design performed marginally better, so you choose to test it again. You could notice that your test was statistically significant but it still didn\u2019t achieve your goal. In that case, you may decide to conduct another test.<\/p>\n The number one rule of A\/B testing is not to guess. The whole point of testing is using quantifiable data to inform your design decisions. A\/B tests are a waste of time, energy and money if you don\u2019t wait for the test to play out or ignore the results.<\/p>\n The amount you\u2019ll use A\/B testing in your UX design career will depend on where you\u2019re employed. Some companies use A\/B testing most of the time, while others rely primarily on usability testing. Regardless, it\u2019s one of the most useful skills<\/a> to have as a UX designer.<\/p>\n","protected":false},"excerpt":{"rendered":" In this guide, you\u2019ll find out what A\/B testing is, how to conduct one and why UX designers should always A\/B test.\u00a0<\/p>\n","protected":false},"author":24,"featured_media":7249,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[278],"tags":[],"class_list":["post-7248","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-design"],"_links":{"self":[{"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/posts\/7248","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/users\/24"}],"replies":[{"embeddable":true,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/comments?post=7248"}],"version-history":[{"count":5,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/posts\/7248\/revisions"}],"predecessor-version":[{"id":9694,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/posts\/7248\/revisions\/9694"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/media\/7249"}],"wp:attachment":[{"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/media?parent=7248"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/categories?post=7248"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.uxdesigninstitute.com\/blog\/wp-json\/wp\/v2\/tags?post=7248"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}What is A\/B testing?<\/h2>\n
What are the benefits of A\/B testing?<\/h3>\n
Why A\/B test?<\/h3>\n
\n
What to A\/B test<\/h3>\n
\n<\/span>– Copywriting, which includes:<\/p>\n\n
\n– Icons
\n– Buttons, which includes:<\/p>\n\n
\n
\n
\n
How to conduct A\/B testing<\/h2>\n
1. Set a goal<\/h4>\n
\n
2. Determine what to test<\/h4>\n
\n
3. Hypothesise<\/h4>\n
\n
3. Conduct test<\/h4>\n
4. Wait<\/h4>\n
5. Analyse results<\/h4>\n
\n
How to analyse A\/B testing results<\/h2>\n
Whatever you do, don\u2019t guess<\/h4>\n