A/B vs Multivariate Testing: How to Choose the Right One

By Melissa Duko, October 10, 2017

Testing... Testing... 1-2-3. It’s rare to get something perfect the first time (unless you’re Beyonce). That’s why there’s testing. Through trial and error, you’re able to make modifications to fix what’s wrong until you get it right.

For pay per click advertising there are two types of testing digital advertisers can try: A/B and multivariate.

Not sure which one’s right for you? Here’s what you need to know.  

What Is A/B Testing

Also known as split testing, A/B testing is a form of website optimization. It compares two different versions (A and B) of the same thing to see which performs better according to live traffic.

A/B testing is handy for building better creatives. You can test variables like call-to-action (CTA), copy, visuals, targeting, and placement to see which version has a better rate of conversion.

For example, let’s say you want to test your subscriber CTA. You have the original CTA (green), and the modified CTA (red). You assign the original to Version A, and the modified one to Version B. Using live traffic, you compare which page did better.

WordStream CTA Examples.png

Source: WordStream

Hypothetically, let’s say the green button knocks it out of the park. Now you know which CTA version to use on your site.  

What Is Multivariate Testing

Multivariate testing is similar to A/B testing, but compares a higher number of variables at one time. Plus, it helps pinpoint how these variables interact with each other (e.g. headline is clashing with your CTA).

optimoid-multivariative-testing-.png

Source: Mockingfish

When you’re looking to redesign specific elements of your page, multivariate testing can help you target those efforts more efficiently. Typical multivariate tests include testing:

  • Visuals and text.
  • CTA’s and text.
  • Form fills and text.

With each combination, you’re able to see if the conversion rate improves or decreases.

How to Conduct Your Testing

To make your A/B or multivariate testing count, make sure you’re focusing on your biggest pain points. If you aren’t sure exactly what those pain points are, circle back to your buyer persona (e.g. the person you’re targeting).

Once you’ve identified which variables you’re testing, be sure your sample size is significant enough to ensure high statistical power. The larger the sample size, the less likelihood of a false positive result.

fede2b0f9d0bbcf93c54430ea12d9db2.jpg

Source: Pinterest

And there are a variety of A/B testing tools such as Optimizely, Pluralis, and Google Analytics that you can use, too.  

So, Which One Is Right for Me?  

So, the million dollar question: which one is right for me? Well, it depends. Remember, split testing only lets you test one variable at a time, while multivariate uses multiple variables.

If you know that there’s an issue with your CTA button, it makes sense to run a split test. But if you’re planning to overhaul your web page, then multivariate testing makes more sense. Split tests are expensive to run, and here, you would be evaluating several variables that would quickly add up if conducted as individual tests.

Carefully weigh your goals and needs to determine which method will work best for you. And remember, you don’t have to commit to just one. Feel free to switch it up. Happy testing!

New Call-to-action
Melissa Duko

Melissa Duko

Melissa Duko is the Senior Editor and Digital Specialist for eZanga. She brings to her role 11 years of journalism experience and a love of all things pop culture. A graduate of the University of Delaware, she holds a Bachelor of Arts in English, concentration business and technical writing, minor Art History. She also has a Master of Science in professional writing for the public and private sector from Towson University. She isn’t afraid to admit that her love for Starbucks is at gold member status. (Since 2011!) And her penchant for retaining pop culture trivia means she knows what "rickrolling" is and isn’t afraid to use it. More Articles by Melissa Duko