One of the main goals for any e-commerce business is to get visitors to make a purchase or a conversion. Many businesses think driving traffic to their website is all they need to increase conversions and revenue. While that’s a key part of the conversion process, it’s only the first step.
Conversion doesn’t happen by itself; it’s achieved through optimization. How do you determine what to optimize on your website though? That’s where conversion rate optimization (CRO) comes in.
CRO is the process of optimizing your website so visitors are more likely to complete a desired action. It allows you to identify, test, and implement optimizations that will have the most impact for your business. Let’s look at the steps involved in CRO and how you can apply them to your website.
Research and Discovery: What Needs Improvement?
The first step to CRO is identifying potential areas for improvement and optimization on your website. Start by gathering user data that will inform and guide your testing.
Google Analytics can provide a detailed snapshot of who your users are and their behaviors. Be sure to have the proper goals or e-commerce tracking enabled before gathering data, though. This allows you to measure conversions and more importantly, the results of your CRO tests.
Through Google Analytics, you can see how users are interacting with your website, where they’re dropping off, and which pages could use improvement. Other visual analytics tools such as heat maps can also provide further insight into how users behave on your pages and forms.
While quantitative data is a big part of the research stage, it often only focuses on the “what,” “when,” and “how” of user behavior. Just as important is understanding why users are behaving the way they are. Surveys are a great tool for gathering user feedback. They can help uncover the actual reasons why users are struggling with your website, straight from the users themselves.
Formulate Your Hypothesis
After you’ve collected your data (both quantitative and qualitative), look for common patterns and trends. These will help you decide what areas of your website to focus on. Use these findings to create several hypotheses that outline what you plan to test.
An effective hypothesis contains three key parts: the proposed change, desired effect from the change, and the rationale behind why the change will lead to the desired effect.
For instance, your data shows most customers abandon their carts during the checkout process. So, you want to test if the number of form fields has an effect on completed purchases. An example hypothesis might be: “We believe shortening the billing form will increase purchases by X% because it speeds up the checkout process for customers.”
Testing: How Can We Fix It?
Using your hypothesis (or hypotheses), you’ll run tests on your website to either validate or disprove your hypothesis. There are two main types of optimization testing: A/B testing and multivariate testing. Let’s look at the differences between these testing methods and the common uses for each.
When you run an A/B test, you create two versions of a web page, and split the traffic evenly between each page. For each page version, analyze how visitors interact with the page they’re shown. This allows you to compare the conversion rates and determine which page version is most effective.
Since it only focuses on a few tracked variables, A/B testing is a quick and easy testing method for gathering meaningful results. Both large and small websites can benefit from this method because it doesn’t need a large amount of traffic to run. Keep in mind, A/B testing is most useful for measuring the impact of two to four variables. Any tests with more than four variables will take longer to run.
In multivariate testing, you’re not testing different page versions like with A/B testing; instead, you’re testing different elements on a single page. Multivariate testing allows you to track multiple variables and how these variables interact with one another. By understanding these interactions, you can identify and improve the specific page elements that will benefit your website most.
While this testing method allows you to target your optimization efforts, there are a couple limitations when using multivariate testing. The main limitation is the amount of traffic needed for testing. Multivariate testing involves more variables than A/B testing, so it needs higher traffic to run. You’re also testing more page combinations in a multivariate test, which can make tests more complicated to run.
Which testing method should you use? This often depends on your unique website goals. Whether you choose to do A/B testing, multivariate testing, or both, make sure your tests ultimately align with your specific website and page traffic.
Analysis: What Should We Do Now?
You’ve done the research, conducted optimization tests, and now it’s finally time to put everything together. At the analysis stage, you’ll review your test results and decide how to optimize your website.
First, ask yourself, “Was my hypothesis correct?” If yes, that’s great! The next step is to plan how to implement these changes to your website. Work with engineering, web development, and design teams to determine the implementation cost and if the projected revenue outweighs the cost.
If your hypothesis was incorrect, don’t dismiss the test results and go back to the research stage immediately. Instead, look for takeaways and learning opportunities from your tests. Analyze the data for additional insights that weren’t accounted for during the initial research.
In most cases, you may not have to create a whole new hypothesis but simply revise the current one with your new test findings and insights. After revising your hypothesis, revisit testing and repeat the previous steps.
It’s important to remember that CRO isn’t a “once and done” task but rather an ongoing process. There are always areas of improvement for your website and overall business. Having a well-defined CRO process helps you identify and implement optimization efforts with the biggest impact on your website.