Split Test Without the Splitting Headache

Marketing is like a slot machine. Think of the millions of combinations you can end up with when you pull that lever or push that button — and how random it is that you’ll get a great return when you do. But that’s how it is when you treat marketing like gambling, as opposed to a science. Your gut can only take you so far. At some point, you need to look at what the raw data is telling you and split test your marketing.

split test

Think of split testing (aka A/B testing) as a scientific experiment, like the ones you used to conduct back in high school science class. You have a hypothesis that one particular headline, design, and call to action combination will give you a great conversion rate or spur people to take action.

And like one of those experiments in high school, you use the first test as a control, and change one variable. Then, you test the new option to see how that one tweak affects your results over a set period of time for both.

If the new one is better, keep the change and try something else! Your goal is to see what gives you the best possible conversion rate, so you never really have a set endpoint for testing. And the thing about marketing is you never know what’s actually going to work ahead of time — just like you’ll never know what the next pull of the slot machine lever will bring. Only data holds the answer.

Split Test Variations in Your Design

What are some design elements you can tweak to see what gets you a better result in your landing page split test? There are countless options you can work with, but we’ve listed a few to get you started. Consider these variations:split test

  • Color scheme: Are you going for bold, muted, flashy, soothing?
  • Font style: Is it adventurous, serious, eye-catching?
  • Call-to-action button: What size, shape, and color is it? Does it include effects like a drop shadow?
  • Photo image, illustration, or video: What combination of media works best?
  • Casual versus professional: What kind of audience are you appealing to? Does it work better to have a more casual design theme or keep it strictly professional?

You can create variations of one of these elements to see which gets you a better response over a predetermined time frame. Visual factors like these can subconsciously influence how someone reacts to your landing page.

Considering Psychology in Your Testing

Of course, there’s more to marketing than design. Psychology plays a big part in how you craft a landing page as well. Here are some psychological factors you should consider:

  • The message or value proposition of the piece: What’s your offer and unique sales proposition? Which angle are you taking in appealing to the audience?
  • The body text: Is there too much text? Too little? Can you tweak the subheads? Should the body text be more emotional, or more logical?
  • The features and benefits you highlight: You can overwhelm the reader with too many features and benefits, but are the ones you chose to highlight the most impactful? See if some other options are better.
  • The lead gen form: Where is the lead generation form located on the page? Are you asking just for a first name and email address in the form’s fields, or a full name, email, phone number, and more details?
  • The wording of your call to action: “Buy now”? “Click here for your FREE gift”? “YES, I want to take control of my life”? Try variations on the call-to-action wording to see which phrasing gets your audience to click.

split testNow, remember that science experiment, where you change only one variable? This makes sense for isolating the reason why one page performed better than another. But it’s very subtle, and you may not be able to wait months or years for the split testing to play out.

Depending on your goals, you may find it better to introduce a significant number of differences into your landing pages early on. In doing so, you’ll probably see a more dramatic disparity in results. But it’ll also be harder to point to the cause of that improvement, and it can be a shock for return visitors. This is the trade-off you have to contend with.

Final Tips for Your Next Split Test

From the beginning of any new campaign, it’s crucial that you have something specific in mind to test. For example, if you look at your current landing page and observe that the call to action doesn’t appear until the bottom of the page, you may consider adding a large red button above the fold that lets visitors take action right away. Then you can discover whether or not that change is a net positive for conversions.

Keep in mind, you’ll need at least 100 unique visitors per landing page in order to get test results that aren’t skewed by a small sample size. You should also make sure the test lasts at least a week (depending on your site traffic) for the same reason.split test

While you’re running a test, it’s best not to have other tests going at the same time. You can certainly have three or four variations of the same element in a test, such as a handful of different headlines being tested. But even then, you’ll want a bigger sample size and a longer test to make sure you have enough data.

Running multiple tests simultaneously for different elements — such as headlines, calls to action, and lead gen forms — is too complicated for most people to draw any useful conclusions from. There’s nothing wrong with keeping it simple!

Ultimately, it’s counterproductive to stress about split testing too much. Yes, it matters, and it truly can improve your bottom line. But a razor sharp focus on split testing can cause you to lose sight of the big picture.

Instead, treat split testing as what it is: a fascinating marriage of creativity and data analysis. Let your imagination run wild, then, let the data bring you back down to earth.

For examples of split testing in action, check out Behave.org. And to implement your split test, we’ve had good results with Unbounce and ClickFunnels!

 

, , , , , , , , ,

Comments are closed.
↓
%d bloggers like this: