Split Test Myths And Statistics

  • Sharebar

Several years back I created Easy Split Test for marketers who want to improve their sales copy.

A split test is where you send one page’s traffic to two pieces of copy and calculate which page converts best. Afterwards you stop showing the loser and begin a new test with the winner and a new slightly modified version of it.

One myth is that there must only be ONE difference between the two pieces of copy to have a valid test. That’s not true at all, and marketers have been doing the exact opposite for decades.

Businesses often enlist copywriters to write copy to challenge their control piece.

They may have copy that has been running for years and converting a solid 2%, but they want to see if that copy can be beaten. So a copywriter writes an entirely different ad for the same product to see which ad converts better.

That’s a valid split test.

The only problem that may arrive is in the conclusions you may draw from the test.

If the control ad has a red headline and the challenger has a blue headline, you could not conclude which headline color converts best from the test unless that was the ONLY difference between the two ads.

So if the the control and the challenger are completely different ads, you will not find out the answer to small questions like which order button, background or font works best. But you CAN find out which ad converts best – and that’s useful information.

Once you know which ad converts best, you can begin slightly modifying the winner and testing it against itself to find out the smaller questions – like which order button converts best.

Another mistake is thinking you need a certain number of actions before you stop a test.

That can be a good rule of thumb, especially for offline marketers who struggle using a computer, but now there are calculators that can do the complicated mathematical formulas for you.

One such complicated mathematical formula is the Chi Square test. This test can predict which version will be the winner from surprisingly low amounts of data, so it doesn’t take as long to run a test. (Which is great when you want to optimize a sales letter fast.)

Easy Split Test has the Chi Square test programmed in to it, so you’ll be notified when your test hits 95% confidence.

And you can search google for “Chi Square calculator” to discover the statistical confidence of any test you’re currently running.

Either way, the marketers who are #WINNING are split testing. Go ready spaghetti and get to work!

Leave a Reply