Let’s say we have two florists. One who is afraid of changing anything on his website, in fear of negatively influencing the good/decent results that he’s been getting. And we have another florist who, unlike his counterpart, is relentless in optimizing his web presence, and is not satisfied until he achieves great results. One florist is complacent. The other diligently tests out the best ways to get the desired results.
Split testing (or A/B testing) is far from a marketing gimmick. It is a marketer’s best “Scientific Method” of achieving the best results. You are literally testing different elements, putting them against each other, and seeing what actually sticks and works best for your website. A few small tweaks can yield unexpectedly great results.
The good thing is that, like many other marketing practices, split testing can be done easily, without requiring too much resources or effort on your part. But it requires a deeper understanding of
A) Your goals.
B) Different ways you can achieve these goals
C) Implementing any changes to have success
Defining the goal
It’s not recommended that you split test every single element on your website, because in many cases it’s not necessary. But if you are trying to, say, get more people to purchase your product offerings, there are many elements on your website that you can test. The same goes if you are trying to get more people to sign up to your email list, or if you’re trying to get more people to click on your blog posts. Anything goes. However, it is important to remember to use split testing for things that you want to improve upon. You shouldn’t split test randomly.
Below is a list of different elements that you can split test.
Call to Actions
The CTA is one of the most important elements that you should split test. Things like the color of the CTA button, the wording and language used, size, placement and layout. Don’t be afraid to test these elements.
Website Layout
Switch the style of the website. Change the layout drastically or do a few simple changes. Things like headers, banners, sidebar placements, navigation bars can be altered.
Web Copywriting and headlines
This is huge. A changed headline can make a huge difference. And you should think of your copywriting as something that can be improved. Put your copy to the test and see what results you get.
Images on the web pages
This is another visual element that shouldn’t be overlooked. Some images will perform better than others. Whether it’s an image in the foreground or in the background, don’t be afraid to test this element.
Product pricing
Test what price points work best within your market. You should test pricing strategies like loss leaders, premium packages, money-back guarantees. In many cases, you will find that certain price points perform better than others. This is crucial information to have if you are selling flowers online.
The checkout page
The smoothness of the checkout page can make or break the sale. You need to test different variations of your checkout page. The checkout is the last step a visitor will take when they are buying your product, so you need to make sure that the process is as close to perfect as possible.
The best practices for split testing
There are a few simple things to remember when you are doing a split test.
The two versions of your website should be tested simultaneously. You will usually test the existing design (often called the Control) versus the modified design (the Variation). You should try to test both versions simultaneously, because that will give you the best handle knowing which one is superior. Split the traffic between the two and see what version wins out.
Talk to your designer or webmaster to run the split test. If you’d rather do it yourself, use a split testing tool (like Optimizely, for example).
Don’t give up on the test prematurely. Split testing can take time. Rushing the process of collecting meaningful data is ill-advised, and will cost you. If you’re not sure when to stop the split test, use an online calculator (like this one). That tool will give you an idea when it’s time to stop the test.
Don’t trust your instincts, trust your data. Acquiring meaningful data so that you can make improvements is the only reason why you would run a split test. If you choose to ignore the data you acquire, and instead elect to “trust your gut”, then your website conversions will not improve. Even if the results from the test seem ludicrous, trust them over your biased instincts. Those “surprise” tweaks can make an incredible difference.