4 Keys to successful A/B testing
This is a guest post by Ryan Gould, VP of Strategy and Marketing at Elevation Marketing.
Broadly speaking, marketers implement their strategies with a particular goal: to generate more revenue. To increase conversions and boost that revenue requires art in the form of creative ad campaigns that speak to a target audience. Only science, however, can tell a marketer which of their actions is leading to measurable improvements.
This science comes in the form of A/B testing.
Just like in the lab, A/B testing is a controlled experiment that, when done right, establishes causality between actions and results. Yet marketers often don’t get it right for the reason that A/B testing is only a sample of broader audience groups, and there is an element of uncertainty in the natural variance of these groups.
How does one overcome that uncertainty? Basic A/B testing looks at the differences between the behavior of two groups of people regarding key performance indicators. Statistical significance then creates the mathematical model used in the testing that reveals the validity of the KPI data. For businesses and data-driven marketers, this can be a gold mine.
In fact, major brands swear by the process. Companies like Amazon and Google run thousands of A/B tests a month. This makes particular sense when you consider that 1 out of every 8 A/B tests drives significant change for an organization. Moreover, 60% of companies that do A/B testing believe it is valuable for conversion rate optimization.
So with that in mind, here are some essential strategies marketers should use when putting together an A/B test that will accurately reveal causality and statistical significance.
1: Focus on the audience
The first step to A/B testing is just like the first step of any other marketing strategy: consider your target audience. Who are you trying to reach? What are their needs? Learning this is the starting point in the process of testing your marketing hypothesis.
One quick and effective way to do so is through surveys. You can incorporate surveys in your paid advertising but it’s more effective to simply add a survey form in the sidebar of your website. There are plenty of online survey tools that make it a breeze to do this, and effective questions are as basic and time-tested as “how can we improve your experience?”
Other subjective questions include:
- How user-friendly is our website?
- How engaging is the logo?
- What were your concerns, if any, about shopping with us?
- What prevented you from completing your purchase?
And so on. You can also mine direct customer feedback through traditional methods like email follow-ups, focus groups, and phone interviews.
2: Don’t make assumptions
While not ignoring consensus wisdom and other brands’ successful ad strategies, neither should you assume those recipes will always work in your kitchen. In other words, don’t assume anything.
For example, the general consensus is that CTA buttons should typically be red, as this is a passion color. It may very well be the case that red should be the color for your CTA, but don’t assume that. Maybe your brand logo is yellow and thus your potential customers will respond better to a yellow CTA button, as it’s in keeping with the theme. Test to find out the best choice.
Even though you shouldn’t assume, you can use many assumptions to your advantage. You can create a system taking into account general assumptions and narrow them down to only those that most apply to your brand. For example, here’s what we know about web surfing:
- The average person spends about 45 seconds on a web page
- People don’t read blocks of text; they scan
- Bold colors grab attention
- Website colors should match brand colors
Taking all the assumptions into account will help you narrow down your ideas, whether it’s what color your CTA should be or what your brand logo should look like.
3: Define success metrics
Success is an abstract notion. To get there, you define what kind of success you’re looking for. Knowing how to measure success before testing is the way to do this. Choose specific success metrics that are narrow in their definition. Here are three such metrics that are invaluable:
- Bounce rate—considers how many people arrive on your landing page and then quickly leave
- Exit rate—people who make it beyond the landing page but then leave after arriving at a certain other page
- Engagement metrics—considers the average time visitors spend on your site and the average number of visitors to a certain page
Focusing on metrics such as these will serve you well in the long-run in what is essentially a short-term goal for increased conversions.
4: Document everything
Document all test results. Don’t let anything slip through the cracks. Believe it or not, there are businesses who don’t keep a detailed log of test results. Even many who don’t take the extra (and invaluable) step of building a library of test results. Doing so ensures you not only don’t repeat the same tests but your library data will make future site redesigns a breeze.
Now that you understand the fundamentals of correct A/B testing, the question remains how often should you do it, and where should you focus your efforts. Each organization is different, but stats say that 71% of companies run two or more A/B tests per month. The majority perform these tests on websites, landing pages, email marketing, and paid search campaigns. But don’t test everything—instead focus on tests most relevant to your website and which can produce higher conversion rates for your most important metrics.
Vice President of Strategy and Marketing Services
From legacy Fortune 100 institutions to inventive start-ups, Ryan brings extensive experience with a wide range of B2B clients. He skillfully architects and manages the delivery of integrated marketing programs, and believes strongly in strategy, not just tactics, that effectively aligns sales and marketing teams within organizations.