A / B testing is like the scientific method of websites. It allows you to run experiments to determine what works on your site and what does not work, and through continued testing, you can significantly increase your revenue and conversion rate.
What is A / B testing?
A / B testing is very simple in concept. You have your current website “A”; which acts as a control for this experiment. You then make a small change to your website – add a button here, change some colors, rearrange the layout, etc. This is version B, your “hypothesis”.
You set both sides and then direct a percentage of the site’s traffic to version B. This can be a small number or anywhere up to 50%. You can even run multiple tests, with a version C, D and so on.
You let your experiment run for a certain amount of time and at the end you can use analysis software to measure how well each version performs. Usually everything can be done under one software package, and it will be very easy to install. What you are usually looking for is a higher conversion rate (the percentage of users who buy your product or achieve your goal).
You may find that version B with an extra button here or there works a little better than version A. If so, you can replace version A with version B, and now your website is a little better.
A big pitfall for A / B testing is what is called local maximum. That’s when you A / B tested everything you could think of and came up with the best version of your original website, optimized as much as possible. But maybe your website does not need to be improved. What if it needs something completely different? This is called global maximum, as this chart from Optimizely shows quite nicely:
A / B testing will not magically make your website amazing – it’s just part of the development loop. You still have analytics results, make assumptions about what you can change to make your site better, and implement those changes. But being able to measure exactly how changes affect your conversion rate and other statistics is a big part of this process.
A / B testing is not just about marketing. You can run A / B tests to measure all types of dependent variables; For example, you can run a test with and without a CDN to determine how it affects the loading speed of your site and ultimately your bounce rate.
What is a landing page?
A landing page is a highly optimized page designed specifically for users to “land” on when they click on an ad for you or find you in the search results. It differs from the website because it is much more streamlined and focused on. This is a large area where A / B testing really shines, because you want your landing page to have a very high conversion rate.
Shopify’s main page, for example, has a lot of content, a top menu with many pages, a login button and a button to start your free trial period.
But their landing page is very different. Instead, the menus are gone, the login button is gone. It is assumed that the user will come here specifically to be persuaded to start a free trial period, as they would not have clicked on an ad if they were already a customer.
The only thing on this site is a minimal amount of marketing information and taglines, along with a call to action above and below the middle. There is no other way to distract the user, and the only way to leave this site is to click “Start Free Trial.”
There are plenty of services on the market to help you build landing pages. Unbounce, Leadpages, and Instapage all have drag-and-drop editors along with pre-built templates to create a landing page painlessly. If the tool does not have built-in A / B testing, you can always use a stand-alone analysis service to run tests. For WordPress, there are free plugins like Elementor that let you build built-in landing plugins. For companies, there are services like Optimizely with huge analysis platforms behind them
How to run A / B tests
If you want to run A / B testing, you want to use an analysis service to make it easier. The easiest service to use is Google Analytics, which is completely free to use and supports A / B testing. You probably want Google Analytics on your site anyway, even if you do not plan to do much testing.
Google Analytics calls A / B testing “experiments”, which you will find in the sidebar under Behavior. You can choose which goal you want to measure (bounces, page views, conversion rate, purchases) and forward a percentage of traffic to the experiment page (page B).
If you were to pay for a service, you could get many interesting features. Crazy Egg has the ability to generate heat maps on your site that show where users are interested, which can give you an insight into how to make changes. It also has built-in A / B test support.
Most analysis software has A / B testing in one way or another. You can read our guide to analysis to learn more.
There are some other areas you can run A / B tests. Facebook allows A / B testing on ads, so-called split testing, which can help you get the most out of it. You can test ads on different audiences, different placements and different types of ads. OptinMonster runs A / B tests specifically on registration forms, with the intention of maximizing lead generation.