Conversion Optimization: Split Test Best Practices

Ari Gillamconversion optimizationLeave a Comment

conversion optimization

conversion optimization
Conversion optimization is the process of increasing the percentage of site visitors who complete a specific action. Whether you’re generating leads or driving revenue, there is always room for improvement. Optimizing your conversion rate requires a disciplined testing strategy that identifies what’s working and what isn’t. In this article, we’re going to lay out a foundation for planning and executing a split test that ensures intelligent insight.
The key to a successful test is evolution, not revolution. If you test multiple variables at the same time without the proper structure in place, it’s impossible to identify the cause of your test results.
A/B split testing is one of the most basic and common methods of conversion optimization, but even experienced marketers make simple mistakes that distort the results. In this article, we’re going to lay out a foundation for planning and executing a split test that ensures intelligent insight.

STEP 1: Understand Your Conversion Funnel

conversion funnel
The first step to conversion optimization is understanding how visitors interact with your website. Review your site, email, social or advertising analytics to map your visitor’s journey and identify the stages of your conversion funnel with the largest abandonment rates.

Campaign: Email Newsletter
Associated Metrics: Delivery Rate, Open Rate, Read Rate, Unsubscribe Rate, Click Rate, Conversion Rate

Existing data identifies areas in need of optimization and it also provides baseline conversion rates, so you can clearly track the effect of any changes.

STEP 2: Defining Your Conversion Optimization Goals

conversion optimization goals
While A/B tests are focused and simplistic in nature, it’s important to carefully document your goals. Keep your goals simple and well-documented.

Your conversion goals should be quantifiable, and solve specific conversion problems.

Problem: low email open rates
Goal: increase email open rates

Identifying and recording your conversion problems will help track success and is the first step in theorizing about the cause of the problem.

STEP 3: Predict the Test’s Outcome

optimization hypothesis
A clear hypothesis defines why you believe a problem occurs. Explicitly documenting your hypothesis helps facilitate a fluid exchange of knowledge. Now, not only are you – the split test creator – aware of the details of the testing process, but you can easily inform and catch-up any member of your team.

Your hypothesis must be specific, predictive and quantifiable.

Problem: low email open rates
Goal: increase email open rates
Hypothesis: Using a company name for our email sender name could be making our emails look automated, irrelevant and unimportant. Including a human sender name may make our emails appear more personalized.

There is a clear predictive outcome: a human sender name along with company name will outperform just the company name. It is quantifiable: email open rates can be tracked and analyzed. And most importantly, it is specific.

Specificity in an A/B test means there is only one variable tested at a time. While this may seem intuitive, this is a fundamental aspect of an A/B test that can get lost, especially when the test becomes more conceptual. Isolating one element is the best way to accurately assess the test and optimize conversions.

STEP 4: Determine Statistical Significance

conversion optimization significance
Statistical significance is the probability that your test results are consistent and reliable. Calculating statistical significance is a rather involved process that essentially compares your data to the standard deviation. We prefer to use KISSmetrics’ online calculator.

As a general rule, the more data, the higher the statistical significance of your conversion optimization test. More data comes from a larger sample size or longer test duration – depending on your site and traffic- but make sure you have enough data before analyzing data.

Do not be quick to jump the gun and implement changes based on the way a test is looking like it will turn out. Remember to mind the test length you have set – it can be tempting to check stats earlier, but it’s very important to wait before determining a winner.

We like to reach a 95% statistical certainty before concluding the test. This means that there is only a 5% probability that the test results are an anomaly.

STEP 5: Analyze and Optimize

split test optimization
After your test has run its course, it’s time to analyze the data.

If there is a statistically significant winner, take that winner into a second round of testing. Or, if you feel confident that that variable is an element that will continue to produce the most optimal results, include that winning element and run a brand new A/B test.

Keep Testing

A/B testing is a low-cost way to continuously improve your metrics. A/B tests are data-driven proof behind any changes you are looking to implement. Follow this 5 step template for an A/B test and your tests will be clean, carefully tracked and easily analyzed. The implementation process is easy when you lay out a strong foundation!

If you’re interested in increasing conversions and optimizing your site’s performance, contact UpSellit’s conversion optimization experts!

Leave a Reply

Your email address will not be published. Required fields are marked *