What’s stopping most startup CEOs from achieving their financial goals? It’s as simple as a lack of conversion testing. Split testing or A/B testing is a straightforward and effective way to tap into the mindset of your audience and learn what’s making them buy or bounce.
Many testers are relying primarily on guesswork when running split tests, and that can lead to skewed results and misinformation. Don’t waste your time! Follow these best practices for split testing your web pages and give your conversion rate the boost it deserves.
The Basics of Split Testing Your Web Pages
If you’re unfamiliar with how split testing works, don’t worry—it’s not an overly technical concept. Simply put, split testing happens when you create multiple versions of a page on your website and test them against one another in terms of traffic, conversion and other statistics. The test allows you to determine which type of page is more successful for you, both now and in the future.
How to Set up a Split Test
1. Choose your test subject. The first thing you need to do is select a page (or a set of pages) for the test. It’s best to choose the ones that are integral to your sales funnel or filter.
2. Examine the existing data. Use specialized analytics software such as heat mapping or click counting to determine how the page is currently performing. You should focus on understanding where people are looking, scrolling and clicking on this page as is.
3. Decide what elements to test. Based on your analysis of the page’s current performance, select certain elements to test within that page. You may choose text size, vocabulary, color, or a new design—any element that seems vital to how the page performs.
4. Create the test pages. Talk to your web developer and create several variations to the original test page.
5. Split the traffic and run your test! Your web developer will be able to work with specialized A/B testing software such as VWO or Optimizely to divert page traffic among the set of test pages. Built-in or secondary analytics software will collect the data you need to gauge their performance.
Now all you need to do is compare data and make some decisions.
Top Split Testing Tips
1. Don’t rely on your intuition! Unfortunately, this is where most people start with their split tests, possibly due to a lack of understanding of the available data. Remember to examine your existing page fully before setting out to create test pages! Make your changes based on data-supported hypotheses.
Don’t guess. Test. Put your hypotheses in this format to help keep your trial in check:
I believe that making this change will affect this audience in this way because of these analytics.
For example, “I believe that making the top banner larger will affect the mid-funnel traffic positively because the heat-map data shows that visitors interact more with the larger fonts on the page.”
2. Use familiar, personal words in the website copy. Sales writers agree that writing in the second person helps get audience members engaged in the copy since they feel you are addressing them personally. Simultaneously, using personal words such as “I” or “My” on website buttons and links makes your audience feel like they are personally invested in the content.
3. Choose test pages that genuinely affect your bottom line. Don’t dredge up a random web page that has no immediate connection to your sales funnel—this is a waste of time. Instead, carefully select high-traffic pages that link directly to product or sales pages. Why? The people who visited these pages are already highly qualified leads since they took the initiative to check out your products.
4. Compile qualitative and quantitative data from your test pages. Just as you compiled and examined qualitative data from your original page, you need to repeat that process for all test pages. This will allow you to compare major changes in audience behavior. Simultaneously, examine the quantitative data via Google Analytics, Clicktale or any other specialized software package. This information will show you real traffic numbers as they relate to each page.
Pay special attention to the bounce rate on all pages, and compare any changes to the information collected via heat mapping, user surveys and other similar tools.
Your best bet for interpreting the test and coming up with great solutions is a combination of qualitative and quantitative data.
5. Cut down your page response time. Nobody likes to wait for a web page to load. If your site is taking too long to perform, people will go elsewhere. It’s worthwhile to experiment with low-data test pages to find out whether your audience is being affected by poor page response times.
6. Put your call to action in a heated part of the page! You’ve poured over the heat maps and seen screenshots of your web page being scrolled and clicked. Now, use that info to select the best spot on the page for your call to action. This is the most important part of the page, so make sure to test it thoroughly.
Think Your Pages Are Already Converting as Well as They Can? Think Again!
Split testing will inevitably show you chinks in your website’s armor. Can you afford to sit back and assume the best? Probably not. Up to 97 percent of website visitors are actually not interested in your product, and they aren’t going to buy no matter what you do. That statistic itself should motivate you to streamline your website’s filter, optimize your pages and make elemental changes that will positively affect your conversion rate. After all, isn’t that the point of an online business?