5 A/B tests every web designer should build into their process
A new motto I’ve adopted as a freelance service provider is to give my clients “the Disney experience.” I’m all about delivering great results in ways that make my clients exceedingly happy. Happy clients pave the way for a healthy business, and it often doesn’t take a whole lot to make your client experience a fantastic one.
If this is something you want to do for your clients, too, then you need to consider adding some basic A/B testing to your web design process. Some simple optimization tests that you conduct before fully handing over the reins can go a long way toward ensuring your design is functional in addition to beautiful. It’s also an easy value-add to execute, and it will significantly improve the experience for your clients as well.
Why split tests are important
Split tests tell you what works and what doesn’t in your design. By testing different design variables, you’ll find out what types of things will convert better. Conversion is a big factor in determining how effective your design is, so it’s critical to do some testing and optimize as much as possible. Not doing so is leaving money on the table, and no one wants that…least of all your clients.
Plus, many designers won’t test at all. By making split tests part of your process, you’re going above and beyond your competitors. Tested design is proven design, and your clients will be thrilled. You’ll also have data to back up your design choices. Everybody has thoughts and opinions about what should happen in a design, but being able to say “actually, we tested the white background against the black background and white performed significantly better” can take care of the opinionators without giving anybody headaches.
So what should you test? These five elements are a good place to start:
1. The headline
The headline on a website is one of the most critical elements of any design — just ask any copywriter. Placement, typography, and copy all factor into creating an effective headline, and these are all things that are relatively easy to test.
2. Any and all CTAs
The call to action on any particular page or site is where the magic happens. It’s where site visitors will do what you want them to do, or not. Conversions are a key metric to track, and CTAs are what drive conversions. It’s absolutely critical to test CTAs.
You can test for placement, colors and typography, copy, and even test for the specific action you want taken (for example, entering an email address over visiting a specific page). Sub-tests can also be helpful. What types of opt-ins capture the most information, a coupon or a tip sheet?
One thing to keep in mind is that you need to be careful when you’re testing multiple CTAs at once. Be sure you can keep track of which CTA is leading to the conversion. If you’re just beginning to test, you may want to test only one CTA at a time, at least on any particular page. That way you’ll have clearer insight into what could be causing the conversion rates to be what they are.
So for example, you may want to test a pop-up offer and compare how well a coupon does against a tip sheet. Once you have a good sense of which gets more opt-ins, you can test for placement of that particular opt-in on the page. Testing for all variations at once may lead to ambiguous results, so start with one element at a time.
3. Page layouts
Think about how a page layout can guide a user through a particular process. If you’re dealing with an eCommerce site, test different ways to present the products. What happens when you add or remove elements? Do suggested items do anything to affect sales? You may even want to test the product descriptions themselves to see what garners the highest purchase response.
For other types of sites, you may want to experiment with a homepage that includes all the key information above the fold and see how it does against something with a powerful hero that’s got minimal content. Will people really scroll? (And this goes back to testing your headline, too!)
4. Navigation menus
Not all nav bars are made equal. What types of navigation do this site’s visitors use the most? How does a secondary nav bar affect things? What about a robust footer — how do users respond to that?
5. The 404 page
It seems like an odd thing to test, but this could be an interesting experiment to run in the background. Come up with a couple of different 404 pages to test, and see which one increases conversions, time on site, pageviews, or anything else. To get started, try one 404 page with suggested links and another one with a search bar.
Some pointers for split-testing your designs
There are split tests, and then there are multivariate tests. Generally speaking, I (personally) will stick with split tests. I might have more than two variations of something that i’m testing at any given time, but I’ll just be testing that one element (like a headline or a CTA). It’s certainly possible to test two different elements on one page (for example, testing the headline AND the CTA simultaneously) but that goes to a level of complexity that I don’t necessarily want to reach.
I’ve had good experiences split-testing two completely different tests, however — like running a test for a headline on the homepage and another test for a content upgrade within a blog post, or testing a blog post welcome mat and a 404 page simultaneously. The two tests are in universes that are far enough apart that they aren’t going to interfere with each other very much.
What new split test are you going to run today?