This article will explain how to do split testing on your content to improve your conversion rates. Awhile back I was doing a fair amount of PPC on Google, it was cheap to buy “buyers traffic” and you could send them to pretty much any page you wanted. Since then things have gotten much tougher, from restrictions on ad and landing page content to CPC prices heading sky high, but the basic formula for profitable internet marketing has remained essentially the same at it’s most basic form…
Put up a sales page, buy some traffic, make enough sales to pay for the traffic, reinvest the profit!
It’s easy enough to understand the theory, however it’s (only) a bit more complicated when you get down to brass tacks. Side note: Watch my quick video if you’re new to internet marketing, or still not making the profit you want.
What the heck is split/multivariate testing?
One of the areas many people overlook but benefit from learning is called split testing. Split testing, or A/B testing means testing different versions of a page to measure which performs better.
With multivariate testing we’re testing elements within a page. We may, for example, test a red and a yellow “buy now” button to see which one results in a higher click through rate. Then we’d take the winner and do another round of multivariate testing, pitting the winning color against other colors until ultimately one color proved most effective. Now this doesn’t only work with buttons; As you’ll soon learn you should perform testing on a variety of different types of media.
What should you be testing?
In theory, everything!
Consider trying different versions of the ad, headline, image, landing page, offer, bonuses, price point, copy, graphics, text, fonts, and even the types of leads. In that last case (leads) I would use the exact same content as I’m interested in measuring the quality of the leads, make sense?
Also, I’ve had conversion rates triple on an ad by just split testing a few titles, and other tweaks resulted in an increased CTR but less sales. Each testing result gives you more information about how to setup your campaign for maximum profit.
In the button example above I bet you’d find that a yellow “buy now” button often generates a higher CTR. Yep, that’s why you see so many yellow “buy now” buttons (As a side note, I believe paypal’s choice to use this color for their buttons played a large part in this online trend) Now that’s good info, and can increase the profitability of a campaign that’s already generating a positive return.
Split and multivariate testing may sound a bit confusing, but it’s actually very simple if you outline a plan and keep notes. For example, let’s say your goal was to double your CTR from 1% to 2%. I may run a series of split tests on each element of the ad. The more the better, but maybe 20-50 different versions or more, each a little different. As the campaign ran I would be able to see which ads were most effective in terms of raising my CTR. After I was satisfied with the data, I would enable only the campaigns that produced the best result in relation to my goal.
Then just repeat that process with another goal and set of tests.
Now, the amount of traffic needed for split/multivariate testing varies depending on what you are testing. I may drive a few thousand visitors to each version of those buttons but pull an ad that isn’t performing after only 500 impressions. You’ll start to develop a feel for when you have enough data, but also consider what metrics you’re trying to improve when optimizing your campaigns.
There are many variables to split testing, but if you break them up into chunks you can really start to improve key metrics. The key is to start with a goal and a set of tests, then tweak your campaign while measuring each change. Once you’ve found the most profitable combination of creative elements, then rachet up ad spend and let ROI determine the investment.
Leave a comment if you have any questions about split testing or multivariate testing. Remember, you may be in business for yourself, but you’re not in it by yourself.
Comments