The Chatterbox

Conversion Testing 101

Yes, it's finally time for me to start this series on Conversion Testing. Okay, okay... It's well past time to start it and I know it. It's just been very, very hectic! That said let's start digging into it, starting with the Basics.

First, let's start off with what I mean by Conversion Testing.

There are several kinds of tests you --as a webmaster-- can conduct, as well as several different methods you can use. Sometimes you'll see the methods referred to at Split Testing, A/B Testing or even Multivariate Testing. I consider it my job to de-mystify this stuff as much as possible, and frankly I do not think it's necessary for the average person to understand all of the various terms or mathematic formulas that go into conducting conversion testing.

In other words, you shouldn't have to hold a Phd in Mathematics and Statistical Analysis just to conduct a valid test.

We will have to use some common terms though, so that we're all talking about the same thing. I'll point out the terminology I use as I go along, and try to create a little Glossary down the road so that we all stay on the same page. So when you see me referring to Testing or Conversion Testing I'm kind of rolling it all into one big ball.

My goal somewhere down the road here I'll be constructing and releasing a tool that will allow you to conduct tests, as well as know at a glance when the tests are valid and conclusive.

Okay, now that that's out of the way...

What sort of things can you test help you increase your conversions?

A much harder answer here, because you can literally test anything and everything. For instance, you could test the Headline on your landing page to see if one version clicks better with your users. You can also test your order form to see if you're putting too much friction in the path of your visitors. You can test things as small as how a Submit button looks to total redesigns of page layout and content.

Or for that matter you can test your offer, your benefit statements, your price point, etc, etc, etc. Or your can test to see if people coming from one source (eg Google) are buying at a higher or lower clip than other (Yahoo) in order to make decisions on where your PPC emphasis or optimization efforts can give you the most bang for the buck.

Literally, you can test anything.

We're going to take a very simple example to get the ball rolling.

Let's say you have a landing page that has a headline of something along the lines of World's Best Lederhosen Here! and you're not sure that your headline is the best it can be.

So here you have what I'll hereafter call your Preliminary Testing Question. You can ask this Preliminary question several different ways, but they all mean basically the same thing. You could word it: Will another headline perform better and bring me more converions? Or: What will happen to my conversion ratios if I use a different headline? Or: How can I determine if a different headline would increase conversions?

Bottom line, you should always start with a Preliminary Testing Question so that you know where you're at now and what you're trying to determine. Make sure to jot this question down somewhere so that when you refer back to your test weeks or months down the road it'll be clear what you were hoping to test.

Next up, we want to take our Preliminary Question and break it down into something we can use in our test. This may sound goofy, but you need to do it. You need to get yourself a Final Testing Question. Some refer to these two questions as Preliminary and Primary, which has always been too confusing IMO. The words are too close in spelling, so nomral people end up getting them confused.

I use Preliminary and Final, because that's what they are! One is a Preliminary question and the other is a Final question.

There is a pretty easy way to tell when you have a Preliminary Question as opposed to a Final Question. It comes down to how the question is asked. The Preliminary question is more general, while the Final Question is considerably more specific. The easiest way to tell when you have a valid Final Question is if it starts with the word Which. (See, I told ya it was easy.)

99 times out of 100 a question that doesn't start with Which isn't specific enough to be a Final question. 99 times out of 100 a question that starts with Which will be specific enough.

For instance, looking at our above example let's say you've decided to test a few different headlines against your current one. You'll usually want to leave your original in the mix to be your Control, since you (hopefully) already know how well it performs. So your Final Question could be something along the lines of: Which of 4 different headlines will perform the best with regard to Conversions.

Quick rule of thumb... your Final Testing Question should always begin with the word Which. Make sense?

Let's knock off here before this single post turns into a friggin' novel. So a quick review...

• You can (and eventually should) test pretty much anything and everything when trying to improve your Conversions.
• Your first duty when getting ready to start testing is to come up with a Preliminary Question to get the ball rolling.
• Once you have this Preliminary Question get more specific so that you can start constructing your test.

Next up ... Variables, Values and constructing your test.