What if there was a scientific way to validate a marketing tactic on your customers? Further, be able to learn from your customers’ activities without being creepy or annoying? With Split Testing, you can be free to experiment with unique subject lines, different headlines and even different copy to see what resonates with your contacts. In this article, I’ll cover some of the ideas and methodologies of split testing.
At Infusionsoft, we split-test all the time, and you don’t even know it. It’s a process that can turn into a beast if we think it, but for many campaigns, we try different subject lines and content for an initial batch of people and move forward with the winner for the rest. The feedback this process provides is invaluable, as it can help guide you into the mindset of your contacts without being creepy. The results split testing provides is rather mathematical, scientific and helps provide insights on how prospects respond to your messages.
Want to see split testing in action?
Go to our Website, www.infusionsoft.com, and observe the headline and byline. Then, clear your cache and cookies and refresh the page. You should notice that it changes. We track the activity among 180 other variables on our Website with the assistance of Google Site Optimizer and our rockstar marketing team. That, in itself is the process of split-testing.
When most people hear of split testing, they often refer to A/B Split Testing or Multivariate Testing. Simply put, A/B is involves one or two variables and multivariate testing involves several, even hundreds of variables to track. In email, it can be difficult to do multivariate testing, but with patience it can be done; for the most part though, A/B testing is where it’s at for email to indicate clear results.
Split Testing heavily relies to the Scientific Method in obtaining conscientious, reproducible results so time to put on your 6th grade science fair project hats as you charter down the path of split testing your marketing. It’s fun, exciting, and unpredictable!
To make split testing effective, you ought to have a large sample size to work with. Roughly speaking, 500 or more people would be a good enough sample to begin with to make informed decisions. The larger the sample size, the more accurate results will be. The smaller the sample size, the less reliable results are and can become a pain to measure when observing conflicting results.
It’s important to remember why split testing is needed. Many people feel the goal to do split testing is the conversion percentage – I disagree. There are many variables to monitor and make educated judgments at the end of the day. Response, Opens, Clicks (what did they click on), Opt-outs, ISP Spam complaints, ISP-open-distribution and other factors that relate to your business (phone calls, emails, etc.) Conversion is one thing, engagement and interactions are another. Also never forget, we’re testing humans, and people react to marketing differently at different times.
My suggestion is to aim for two to three measurable goals in split testing before doing it. Not having a clear vision or purpose will make the results feel less than worth effort to obtain. Even if you don’t get any bona fide sales-inducing results, at least you will be learning from your subscribers and building your marketing prowess along the way.
The bottom line, split testing works in email, direct mail, phone interactions, one-to-one interactions and of course, Web site content. Try it; one of my friends, Caleb Jennings, once wrote about split-testing real life. How’s that for creativity?
There’s a lot of reading to do on this subject and several solutions out that that enable you to command and control your split testing practices. Stay tuned for more on this topic on the blog … including how you can do split testing with Infusionsoft.
[Photo by blondyimp on Flickr]
