Back to Articles

Why You Should Start Running A/B Tests on Your Website Now (And How to Do It Right)

Published by Andrea Hester on January 16, 2020
Alex Kondratiev H9T723Y Pj Yi Unsplash

A/B testing is easy to do and can be done fairly inexpensively, so not doing it is like leaving free money on the table for your business.

The human mind is surprisingly sensitive to small signals and cues, often in ways that we’re not consciously aware of. As web developers and user experience experts, we’re constantly amazed at how minor tweaks and changes can produce big results for our clients’ websites. A/B testing is easy to do and can be done fairly inexpensively, so not doing it is like leaving free money on the table for your business.

In order to use your precious time wisely and efficiently, there are some best practices when it comes to A/B testing. Follow this process and you can be sure that you’re running tests that are most likely to have the biggest impact on your conversions.

Step 1: Look for problems

Start with your website analytics to identify the most obvious problems, such as:

  • High bounce or abandonment rates
  • Underperforming mobile traffic
  • People using search for things they should find in navigation

You should also compare your own site to your competitors’ sites. When browsing for something specific or going through the checkout process on their site, is their experience easy or hard? How does your site compare?

Screen Shot 2020 02 12 At 11 42 06 Am

Google Analytics will show you bounce rates for your top pages. Look for problems on pages that have higher-than-average rates.

Step 2: Ask "Why?"

After you’ve identified the problem areas, next, think critically about WHY these pages or traffic sources are performing poorly. If the answer isn’t obvious, think about recruiting real users to perform a function on your site and watch them to see where they get hung up. Or, consider partnering with your web developer to use analytics tools like FullStory that allow you to record and watch real users on your site (we do this for many of our clients). It’s sometimes surprising when you take off your “expert” hat and view your site through the eyes of someone visiting for the first time.

Screen Shot 2020 02 12 At 11 52 30 Am

You will learn surprising things about your website by watching real users interact with it, either in person or through analytics tools like FullStory.

Step 3: Form Your Hypothesis

Now that you’ve identified areas for improvement and gathered information about why users aren’t doing what you want them to do, you need to come up with some informed guesses about what changes could impact their behavior. Common hypotheses we work with are:

  • Changing the wording of this call to action button will increase conversions
  • Changing the location/size/color of the button will increase clicks
  • Reducing the number of form fields will increase form submissions
  • Adjusting the size/wording of this text headline will decrease abandonment
  • Reducing the price of this product will increase sales
  • A newly designed header will lead to more engagement
Screen Shot 2020 02 12 At 11 55 51 Am

We recently tested call to action (CTA) button colors for one of our clients to optimize conversions on their pricing page.

Step 4: Set Up Your Experiment

Once you’ve planned your new variation, you need to test it against your hypothesis to see if it rings true. You could just ASK a few users what they think, but we’ve found that what people SAY and what they subconsciously DO are often very different. That’s why it’s helpful to use a tool like Google Optimize (free) or Optimizely (paid) to run your experiment. These apps manage your experiment(s) for you and give you concrete results. All you have to do is input the current page as your “control,” create your variant page (many of these apps have built-in code-free visual editors for minor changes), and tell it what you want it to base the results on (clicks, pageviews, goals, events, etc.). The amount of time you need to run the experiment depends on the amount of traffic your site gets. Be sure to run your experiment long enough to get a good sample size to ensure that your results are reliable.

A/B testing is one of the simplest yet most important ways that you can be continually optimizing and improving your website. If your marketing team is too busy or if you need more advanced development help than your team is capable of, we are happy to pitch in.

Shoot us a message and let us know how we can help.


Sign-up for our monthly newsletter

We'll send you an email once or twice a month with helpful articles, resources, and best practices for busy marketing teams.