Concept Validation
March 6, 2013
Netmark Lays Out Scientific Conversion Rate Optimization Strategy
March 12, 2013

Scientific Conversion Rate Optimization

By John Broadbent


Thanks for tuning in to another episode of Make Your Mark Monday. Today we’re talking about the scientific method of conversion rate optimization. Here at we believe that really you can combine science and strategy to do just about anything, as it comes to internet marketing. And this is probably one of the most fundamental ways, that we’ve seen that happen is, or where the most elementary I guess, way that we’re able to take science and really apply scientific methods, to what we do in internet marketing.

So the first step that we found in conversion rate optimization is to ask a question. And usually the question is what can I do to improve the conversion rate of this particular page? And then we move on and we start to do research. We look at the industry and we look at the particular industry for that website, and we start to check, to see if there are some best practices that others have already tested.

And then we draw some hypothesis. We say, “You know what? I think that if we probably change this, then and then we move on to prediction, then what will happen?” “If we change the heading copy, or if we change this button color, or if we add a form to this spot right here, you know then most likely bounces will go down and conversions will increase.”

Then we move on to the experiment. We actually dive deep into, we run the experiment. But one of the things that we found is test only one piece of the puzzle at a time. My wife and I, we love making pizza and the hardest thing about pizza to get right is the crust. And we found that there are tons of variables that affect the crust. The conversion rate, in this example would probably be the crust, it is one of the hardest pieces to just get right, and so we test things.

But we can’t test more than one thing. If I see what happens when I put less cheese on the pizza, but then also increase the amount of yeast in the dough, then I don’t really know what made the change. And sometimes we change a lot of things. We’ll change the amount of cheese and the amount of sauce and the amount of toppings that go on top, and then we’ll change a bunch of things within the crust itself.

How much water, how much oil, how much yeast, those types of things, but then after we bake it we go “Wow, that was better, but we have no clue what made it better.” And so one of the things is test fewer things and you can get results much, much sooner. Now there is multivariate testing and things like that, but again if you’re dealing with smaller quantities or smaller amounts of traffic on your website, you want to test fewer things, so that you can get your results much, much sooner.

Then what you do is you analyze. You analyze what it is that happened during your test. One of the things that we found is there are by-products of every test. Has visitor flow on your website changed? Are people now doing different things? And every test that we do has some by-products of that test that we can glean from it. So we strongly recommend jumping into doing some deep analysis of the effects of your tests to see what changed overall.

Not just what you’re asking for in the first places is did bounce rate decrease or the prediction there or did conversion rate increase, but what else happened to the website. And then the final step is to repeat this, and do this over and over and over again. One of the things with conversion rate optimization is your job is really never done.

You’re always looking for an increase. We can be pleased with the increase that we’ve received from past tests and from past experiments, but we shouldn’t ever be satisfied as a conversion rate optimization specialist, you should dive in and continue to test and test and test.

Thanks again for tuning in. This Wednesday, we’ll dive into a page of, and we’ll actually show you how to set up some basic experiments using Google Analytics.