Performing A/B testing of the primary communication elements of Inbound Marketing – your website, content, emails and landing pages – is a great way to identify effective (and less effective) ways to communicate your message and convert traffic into leads. The feedback is directly from users, and based on reactions to different designs, copy, offers, calls-to-action, user experiences, etc.
But too often, the results of A/B testing are used to identify how to a product or service should be marketed. A/B testing, however, should follow – not determine – a well-reasoned content strategy and simply used to pinpoint which way of communicating the same message is most effective.
When am I ready for A/B testing?
There’s no perfect time for A/B testing. You could be just launching your content marketing plan, in the midst of executing your current strategy, or somewhere in between. The only time you’re not ready for A/B testing is when you don’t have a plan that outlines what you think is currently happening (your hypothesis) and what you want out of the A/B testing.
Step 1 of the plan: formulate a well-thought hypothesis. If you have some sense of why certain offers or elements aren’t working, you may come up with a hypotheses like one of these:
“Our customers leave our website after they don’t get the product information on the landing page.”
Related Resources from B2C
» Free Webcast: Build Better Products by Identifying and Validating Your Riskiest Assumptions
“Our testimonials are too exaggerated and don’t result into sales.”
“Customers aren’t filling out the form because it’s too long (too personal, etc.).”
Your hypothesis should be detailed enough so that you can start formulating your A and B variations.
A/B testing execution
A complete set of first round analytics that look at all relevant metrics will give you a baseline comparison for your testing and serve as your control. You’ll want to keep this control handy and use it in every test, whether it’s email design, landing page copy, navigation, PPC ads, etc.
What kinds of things can you change?
If you’re committing yourself to A/B testing, there’s no point in stopping at “A” and “B.” Test as many variations as make sense and have potential. You can change just about anything that’s part of your hypothesis. Keep the goal in mind; you’re looking for a way to either confirm or deny your hypothesis when you’re done.
Dan Siroker, formerly of Google, worked on the Obama campaign and became well known for his A/B testing work. He tested seven variables of elements on the splash page alone and evaluated the results. The video is lengthly but when get to his examples at 5:37, you can see how adjusting the user benefits greatly increased the results and donations to the Obama campaign. Keep “user benefit” in mind as you watch this video.
What changes have the greatest results?
After much execution and research, it’s been shown that changing headlines and using better photography or graphic design reported higher successes in A/B testing and adaptations. In our experience, headlines that answer the question, “What’s in it for me?” more attractive than those that simply indicate what the button is for.
Keep in mind statistical significance while you’re evaluating your results. Your data/testing volume needs to be significant enough to draw relatively strong conclusions.
A/B testing can be time-consuming, but you’ll get rewarded with information that helps you tailor your message and other elements in a way that makes them more targeted, powerful and effective.
Sample Approach To A/B Testing
Much like the example in the video above, you need to identify your goals that shape your hypothesis. I’ll demonstrate with an example of something that we currently use in our A/B testing routine.
Hypothesis: A CTA button with the words “download now” will produce higher conversions than our regular text of “FREE Guide.”
Test design: We will take our traditional CTA design which has the text “Download Now” as our control and design a CTA with a button that uses “Free Download.” All other design aspects of our CTAs will remain the same. We will run this test for 6 months using HubSpot’s variation test set up. This feature evenly distributes the CTA to our prospects while leading them to the same landing page. We will evaluate on view-to-click rate. This HubSpot function also offers a clicks-to-submissions rate for evaluation too. Here are our two evaluated CTAs:
During our variation testing period, the control group had a views-to-click rate of 2.2% which includes all of the locations this variation was placed on our site: home page, blog side bar and blog entry footers. Our Variation 2 CTA had a views-to-click ratio of 2.0%. The percentage difference seems minimal but what matters is the decreased chance for conversions. The bottom line of Inbound Marketing is all about closing on more leads by truly meeting their needs with your content.
Results: When we use this to confirm or deny our hypothesis, we draw a conclusion to our A/B testing. As we found in this 6-month CTA study, our visitors are more enticed by reading “FREE Download” rather than “Download now” and our hypothesis is denied.
Learn more about elements like CTAs and their importance to website effectiveness by downloading our “Website Redesign 101: 7 Steps to a Better Site” now!