Every CRO project should begin with a baseline measurement and benchmark of the key metrics. There are many tools available today to marketers for granularly measuring web and mobile traffic and user behavior, but the most common – Google Analytics and Adobe Analytics – should give you all the basic data you need to get started. Avangate has additional tools for tracking and analyzing the shopping cart that can help fill in the gaps, and give you a more complete picture.
When looking at what to measure and benchmark, it’s best to start with the basics, which include:
All of these would give you the core data you need to calculate site, page and cart conversion rate, as well as average order value (AOV). With these in hand, you can now begin the work of optimizing for both transactions and average order value.
The following will take you through the steps and process for setting up and running a successful CRO program. In our model, we’ll focus on the purchase funnel and shopping cart, as they’re the most critical to reducing revenue leakage in the acquisition stage of the digital commerce lifecycle.
As we’ve reviewed, you should now have a thorough analysis and audit of your current site, purchase funnel and shopping cart. Once you know your metrics and have benchmarked your conversion rate and AOV, you can identify gaps and areas that can be tested and improved. Here are some starting points for your investigation:
Knowing what to test and why is part art and part science.
After you’ve reviewed your pages and processes, you should have at least a dozen ideas or hypotheses of how you can improve them.
A good optimization hypothesis contains an IF/THEN statement to show cause and effect and is tied directly to the user’s behavior and the desired outcome. For example, “By adding a security icon on the cart page, the buyer will have less anxiety and, therefore, will complete the order.”
Once you have a list of hypotheses, you have to prioritize them. It is helpful to rank them according to impact and effort–what will move the needle the most with the least amount of effort to implement the test. The hypotheses with the highest impact and lowest effort should form the basis of your initial testing plan.
Conversion Rate Tests
Average Order Value Tests
For instance, 123contactform, a SaaS provider of Web forms and surveys, managed to increase their conversion rate by 63%. They reduced the steps on the checkout page for yearly subscribers with high AOV products from two to one. Simple as that. Speeding up the process gave them a huge jump in conversions and an estimated 12 percent increase in revenue.
The test plan is your guide and blueprint for the testing, analysis and optimizations that come after it. It should include your hypotheses, the specific pages or objects that you are going to test against (the controls), the two to three test variations (depending on your traffic volumes), and your rationale for why those will perform better than the control. As you test, your plan will evolve throughout the process as you gather insights and apply them to the next series of tests.
Like all statistical exercises, conversion rate optimization requires a meaningful data sample in order to obtain a significant finding. Your test plan not only includes what to test, but also must consider how long you need to test for. Calculating your sample size and test duration is an important part of creating your test plan, and may affect how many variations you can “afford’ to test.
Depending on the traffic to your page and/or cart, 15 days (e.g. two 7-day business cycles) is usually sufficient to reach statistical significance with a confidence level of 95%. Any less, and you won’t have the data to support or disprove your hypotheses. Any more, and you’ve lost time and money to put your findings into play. Like everything else with testing, finding the right balance is something that only comes with experience, and more testing. There are best practices, but they are only general rules and are often proved wrong by the exception.
As a general rule, the winning configuration or model should be able to attain a 95% level of confidence before it can be declared permanent. You can get a handy calculator here. That will help ensure that the results you received during the test are repeatable in your production environment.
Remember, too, that testing is an iterative process. Test number two should be built on the results of test one, and so on. With testing, you’re always working in parallel, with your eye on your next text, even before you’ve completed the one in progress. Even when tests fail or have inconclusive results – and they do 25% of the time – they typically provide some insight that can be applied to the next test or live campaign.
Once you’ve conducted your test and determined the winner, it’s time to put them into play. Running the new configuration(s) outside of the test is the best way to confirm the validity of your conclusions, as well as your test plan and methodology. The before and after analysis can also be a humbling experience if the results don’t hold up over time. If they don’t, then you’ll need to consider additional testing. Even if they do, there is no such thing as the perfect shopping cart, page or purchase flow.
Preferences are always changing and evolving, and you’ll never know with certainty what users will react to next. A best practice is to validate tests by repeating them, for example, in another season or with a different customer segment. The only thing you can do is continue to test and look for ways to optimize, which brings us to the last and final step.
You can complete a test, but the testing and optimization is really never over. Not if you want to continue to recover the additional 5-20% of lost revenue. Revenue leakage can occur at anyplace within the Acquisition stage, and there are numerous things you can do, beyond the testing cycle, to source and stem it.
Of those, the most effective, if not also most satisfying, is the recovery of revenue lost during the purchase process. Strategies for recovering lost revenue typically fall into two categories: abandoned cart recovery and unfinished payment recovery.
The most accurate definition of Conversion is “a progression from one state to another.” When the buyer makes the decision to purchase, to upgrade or renew, it is rarely or ever a random, spontaneous act but instead the result of a cumulative process that was possibly days, weeks, months and even years in the making.
The progression from prospect to trial, trial to customer, customer to renewal or upgrade, and renewal to evangelist doesn’t happen by accident. Instead, it typically is the result of careful planning and, in most cases, significant investments. As the customer moves through the commerce lifecycle they are converted from one stage to the next as the result of your efforts. Or they leak away in between the cracks as unwitting casualties during the commerce lifecycle.
CRO, like anything worthwhile, requires patience and experience to master. It is not impossible, but nor is it something that can be done easily, or in a day. What it is really is an ongoing process and commitment, one that with constant experimentation and hypothesis testing can lead you to greater insights and more effective results.
This post is part of a comprehensive eBook on customer acquisition. To learn more about the digital commerce lifecycle, conversion rate optimization, get examples of marketing strategies from digital leaders and much more, check out our Customer Acquisition eBook.