This article originally appeared in WQP July 2020 issue as "Successful Split Testing"
A few weeks ago, my husband asked if I had seen a new series on Netflix. He could not remember the name, so we pulled up the site and started looking for the series in question. Fifteen minutes later, we were about to give up our mindless scrolling when suddenly he remembered the title. We searched and when the show came up, he noticed the image associated with it was completely different than when he first saw the series. It now made sense why he could not find what he was looking for.
Have you ever noticed how these images change on Netflix? Or how your Facebook newsfeed appears just a bit different than it did the day before? Or perhaps how an Amazon product page has a new section or feature?
These giant companies are constantly changing and updating their content, interfaces, graphics and design—and they do not do it haphazardly. They have a strategic approach using a scientific process called split testing.
Historically reserved for fortune 1000 companies, split testing has recently made its way into businesses of all sizes and shapes and for a great reason. It is one of the most powerful ways to improve your core marketing metric (for water treatment companies, this typically means qualified leads) without spending any additional money.
What Is Split Testing?
Split testing is the method of comparing two versions of a webpage (although you can test just about anything that has measurable results, such as Facebook audiences, Google ads and email campaigns) against each other to find out which performs better at a statistically significant level. This form of testing drives an equal number of people to each version and then measures a specific goal to calculate the conversion rate.
For example, if you have 100 people who see one version (Version A) of a webpage and 100 people see the other version (Version B) and five people complete a form on Version A and 10 people complete the form on Version B, then we would know that A has a conversion rate of 5% and B has a conversion rate of 10%.
Perhaps the headline of Version B was different, the form was shorter or testimonials were added. Whatever the change, we now have one version that appears to be converting twice as many people into leads. Then we would simply need to run these numbers through a split-testing calculator, like https://abtestguide.com/calc/, to make sure Version B is truly the winner with at least at a 90% statistical significance. Once we can confirm the winning version, it now becomes the new version (the control). Statistical significance is key, allowing us to be at least 90% certain that, regardless of external variables, Version B will continue to beat Version A.
How does someone come up with what to test? This is the fun part. Coming up with an idea to test requires the use of psychology, data, observation and previous winning strategies to come up with your own scientific hypothesis. A hypothesis is a testable prediction of an outcome of a study.
For example, our hypothesis might state that users will convert significantly more when testimonials are placed right next to the contact form on the webpage.
We would feel comfortable with this hypothesis because, based on psychology and consumer behavior, we know that people are more likely to reach out to a company if there is adequate social proof to back up the company’s claims.
Then we would simply use the existing webpage, which might have testimonials placed somewhere else on the page against the same page with testimonials placed right next to the form where we believe they will be most useful in convincing the person, at the most crucial point during their website visit, to contact us. Then we would use a tool like VWO.com, Google Optimize or Optimizely.com to run our experiment.
How Does This Work For Water Treatment Companies?
A few years ago, I was working with a residential water treatment client to generate more leads online. A lot of people in their area were visiting the page on their website that was focused on water softeners; however, a small percentage of them actually converted by filling out a form or calling the call tracking number on the website.
We tried split testing water softener pages with less product information and more focus on benefits. We also tested special offers and tested shortening the contact form. Nothing seemed to improve the page’s conversion rate.
Then we took a step back and decided to take a different approach. We set up a fun quiz on the page and started asking questions. We asked if the person already had a water softener. If so, what problems were they hoping to solve by visiting the website? If they did not, what was their biggest water challenge to which they were hoping to find a solution? We dug deep and after a few months had a good amount of data.
We quickly noticed that the majority of people in their market already had a water softener and were wanting to upgrade to a system with better water and salt efficiency. They also had softeners from other brands and found it difficult to get service or find parts.
Based on this information, we set up a split test between the original water softener page and a totally new page that addressed these issues. The result was a statistically significant improvement in conversion rate and more high-quality leads for our client. The best part? They did not spend a dime more on advertising.
The Benefits of Testing Culture
The beauty of testing is that it can improve every single step in your marketing funnel. From the radio ad someone hears that triggers them to search online, to the Google ad they click that leads to your site, to the Facebook ad they see after poking around through a few of your web pages, to the landing page they reach from clicking that Facebook ad, to the appointment script that sets a time for a water expert to test their water, to the actual demonstration done in the home… every step can be tested and improved.
Building a culture of testing throughout your entire organization is extremely important. As people learn that it is okay to test ideas, make mistakes, and learn from both wins and losses, your company will benefit from constant improvement.
The process of testing tied with a numerical value provides a tremendous amount of gratification amongst your team when improvements happen. Perhaps even more valuable than the wins are the losses, which teach us what not to do and why.
How Do We Start Testing?
Ask your marketing team how they are using this powerful form of marketing for you and see how you can work together to come up with strategies for testing and improving. Look internally and see where testing can be used in other areas of your organization besides digital marketing. When I worked as a general manager for a water dealership in New Mexico, we had fun testing appointment setting scripts.
Ensure that anything you do test follows the scientific method and reaches at least a 90% statistical significance so you, like Netflix, can be sure the changes you are making are causing lasting positive results and that you understand the cause of those results.
If you have any questions about digital marketing or split testing, Lamplight Digital Media would love to hear from you. Our team has run more than 800 scientific split tests for water treatment companies all over North America and has a vast database of what works in this industry to improve online lead generation.