A/B and Multivariate Testing — Exciting right?  We say, YES!

We’re guessing if you came across this article, you’ve already been making some rounds on tech and design blogs trying to find out what tools are best for your ecommerce site. You’ve read about analytics and search engine optimization, but now you’re seeing the phrase “multivariate testing” get thrown around. All the jargon can make your brain feel overloaded and your eyes start to get very heavy all of a sudden…

It will wake up your analytics!

If you haven’t heard about A/B or multivariate testing yet then pour yourself another cup of coffee and pay attention, because this is something every website can benefit from. You may need to employ assistance in setting it up, but its a great way to get metrics on consumer behavior for your website, so you don’t have to blindly experiment and hope for the best. Multivariate testing helps you make very educated choices about image layout and text content.

It all starts with conversion rates

An important term that goes along with A/B and multivariate testing is conversion rates.  This is simply the rate at which customers go from your front (landing) page all the way through the checkout. If the word conversion reminds you of religion, that is because it is a lot like that. Your website strives to convince people to believe in your products. When they decide to buy your product or service, they are essentially converted from not believing to believing. And because websites are trackable and interactive, you can actually see the point at which customers have decided to disbelieve and move off of your website. When you see a pattern developing on a certain page, you can create tests to improve your conversion rates. When you just want to compare one or two things, it is called A/B testing.  When you are comparing multiple variables at once, it is called multivariate testing.

Informed experiments, rather than wild guesses

For example, say you want people to go from your first page to another page which lists pricing, with a button that says “buy now.”  You notice that people are spending time on your landing page but avoiding that “buy now” button.  So, you set up an experiment.  You make an identical page but change the wording of the button to “view prices.”  Then, you set it up so that each time a new visitor comes, one of the pages appears to them, either the original page or the new one with the new wording.  After a while, you will be able to measure if changing the words of that button improved the conversion rate and brought visitors one click deeper into your site. Not only will you be able to see if it worked, but you will be able to measure the percent of increased conversion. Pretty nifty, right?

So how do you do it?  

Well, the best way for beginners to do so is to hire one of the many new companies that are cropping up which perform just that task.  Just like with any specialized service, do some research before you choose a company, and ask your fellow entrepreneurs for referrals. Once you choose one, they will set up and tag whatever variants you want to test using Javascript. At that point, the sky’s the limit regarding what you want to test: wording, image changes, coupons, trial periods, even color choices and fonts. Just be aware that the more variants you test at once, the harder it can be to know which change is making the difference.