Kirsten W. polled theCLIKK’s private Facebook Group yesterday:

split test theclikk 

One response was “Test it, cause your audience is unique.” I love the enterprising spirit, and I usually agree with the testing sentiment, but I’d be concerned about the particulars here. As a practical matter, should you really test that? More specifically: is it worth your time and effort?

Of course, this raises the bigger question: what’s worth testing and what isn’t?

The shortest answer is “it depends.” If you have a team of Conversion Rate Optimization (CRO) professionals working on nothing but split testing, your definition of “it’s worth split testing” will be far more charitable than the definition you’d get from a small marketing team.

When I worked in marketing at Salesforce.com, for example, we had far more bandwidth for split testing, both in time and in resources—and so we tested damn near everything under the sun.

For the rest of us, two big tips:

1. Test close to the money.

Testing which day of the week to send a content email is pretty frickin’ far from the cash register. While I do believe sending content via email is a fantastic use of many people’s time, testing for the ideal day of the week isn’t likely to yield results (and would be a complicated test to do correctly anyway). Instead, throw a dart at the calendar and begin consistently sending content on that day of the week around the same time.

Take the time you’ll save from testing and use it to test a price point on a product, or the layout of your checkout page, or a money-back guarantee. All of these things are much closer to the money.

2. Test where there’s traffic.

If you don’t have enough traffic and conversions, you can’t test. Not effectively, at least. You still could test without incoming traffic and conversions, but it will take you an infinite amount of forever to get statistically significant results.

How much traffic do you need? I quote this post from Visual Website Optimizer:

“Let’s say, you have an existing conversion rate of 5% and the expected percentage increase in conversions is 10%. If you’re testing 4 variations, here’s how long you will have to wait to get conclusive results (95% confidence level), depending on your website traffic:

 when should you ab test theclikk

So, with a website getting 500 visits per day (which is still nothing to sneeze at), it would take 400 days to get conclusive results. Fuhgeddaboutit!

Does this mean you shouldn’t be watching your numbers and adjusting accordingly based on the data you see? Heck no. Watch your numbers and adjust things smartly. Just try to be choosy about when you’re willing to go down the split-testing rabbit hole. Trust us: it’s dark and very deep down there.

— Russ

Get more from theCLIKK