The obvious truth is that if you’re not testing, you don’t know if you’re doing the best that you can’t. However, if you’re testing and you are relatively new to it, then there are several possible pitfalls that you might face. I would like to talk about one of them today.

Let’s talk about A-B split testing. Just to summarize, if you don’t know what it is, you’re basically testing one item against another. Those two items are different from each other. However, not too different. Usually you’re supposed to test

one variable. For example, a different headline, different sub headline, call to action, order form, etc.

Let me give you a scenario and then explain:

For example, you are testing a pop-up copy. You have two pop-ups with different copy.

After sending very minimal traffic to the website that has the pop up, you go check your stats… and this is what you see:

At the first glance, you might think that the P2 is performing better than P1 and that’s where the pitfalls lies.

While inexperienced split tester would say that P2 is performing better than P1, an experienced split tester would say that there is nothing enough information and the results are not statistically significant.

If you have taken a statistics class, they know what statistical significance means.

However, you’ll need to know what statistical significance is and how to calculate it to find out if your test results are statistically significant because there is a free tool that you can use to figure this out within seconds.

Introducing: Statistical Significance Calculator…

Use this tool to make a better judgment about your test results so that you don’t declare the losing copy in winter, and the winning copy a loser. After all, that would get you anywhere.

Why is this important?

But finding out whether your test results are statistically significant, you can know for sure which version is winning so that you can concentrate on the. Otherwise, you might make a quick decision, too soon, but your test results and instead of improving the performance of your piece, with a quite opposite.

What should I do now that I know this?

Now that you know about this, don’t judge too soon the results of your split-test.

1. Get enough traffic to the web site.

2. Find out if your results are statistically significant by using the free tool above.

3. If your results are statistically significant, then you can keep the winning copy and abandon the loosing one. If your results are not statistically significant, then keep driving traffic until they are.

I hope you see the value in this free tool. Have you ever used anything like this? Leave your comment below…

Split Testing is Worthless, Unless… is a post from: Narek Gabrielian's Blog