Everything you need to know about effective ecommerce A/B testing
It has become a cliché in ecommerce marketing and web design blogs to state that throwing your website out there without instigating a proper testing regime is taking a shot in the dark, throwing darts blindfolded, or riding a bike with flat tires… You get the picture.
Whichever allegory you prefer, the point remains the same. An untested website might perform, but it’s unlikely to perform optimally.
Testing takes the guesswork out of web design. Whereas software developers might rely on a range of test procedures, from ad-hoc software testing to smoke testing, A/B testing has become the industry standard when it comes to conversion rate optimization for websites.
What is ecommerce A/B testing?
A/B testing is a way to compare two versions of a single variable by presenting your audience with both options and comparing results. For the purposes of eCommerce, the most important thing that can be A/B tested is your website, and that’s what we’ll be focusing on throughout this article.
By testing people’s responses to different variations in content, A/B testing is able to determine which of the two is more effective at generating conversions. In eCommerce, conversions typically refer to how many visitors to your website are converted into customers.
Multivariate testing is a variation of A/B testing that tests multiple page elements against a control version to find out which version of each element is optimal.
Multivariate testing can be a powerful tool for ecommerce retailers and many of the aspects that apply to basic A/B testing are also true for multivariate testing. What’s more, testing automation platforms will combine many A/B tests into a comprehensive multivariate test process for you. If you’re wondering what is automated testing? It’s the process of testing software and other tech products to ensure it meets predefined requirements, fully automated.
For simplicity’s sake, we’ll stick to the one term in this post (A/B testing) but keep in mind that multivariate testing and A/B testing have a significant conceptual overlap.
Some people also make a distinction between A/B testing and split testing, but for most website designers, the two terms can be considered interchangeable.
In the world of ecommerce, A/B testing and the applied data analytics it makes possible have been shaping web design for quite some time. For example, it’s thanks to ecommerce A/B testing that we know a slow website hurts conversion efforts.
It’s also because of strict testing regimes that companies like Amazon have been able to fine-tune their page layouts and offer vendors the precise performance data needed to maximise sales.
Advantages of A/B testing for ecommerce
To measure the performance of an online store, retailers use a number of metrics. We’ve listed a handful of the key performance indicators that reveal the effectiveness of your ecommerce website and discuss how A/B testing can help optimise these numbers.
Boosts conversion rates
As you might have already gathered, A/B testing your website is one of the best ways to boost your ecommerce conversion rate.
Experimenting with different content, page layouts, and navigation options can reveal with the help of sales enablement software which version is most effective at turning web traffic into sales.
Reduces bounce rate
Your bounce rate represents the percentage of visitors who enter your site and then leave rather than continuing to view other pages on the same domain.
A/B testing your landing page(s) is one of the best ways to reduce this metric.
Improves return on ad spend (ROAS)
Return on ad spend, or ROAS is one of the most important metrics in eCommerce. It evaluates the effectiveness of a marketing campaign by measuring what the returns are per unit spent on digital ads.
Increases lifetime customer value (LTV)
By helping you create an improved customer experience, A/B testing means that customers are more likely to keep coming back time and time again. Lifetime value measures the total amount a customer spends with a brand from their first purchase onwards.
Loyal customers are great for ecommerce because they provide a steady revenue stream and respond well to targeted marketing campaigns.
Reduces customer acquisition cost
Customer acquisition cost (CAC) is the cost of convincing a customer to purchase a product/service. By reducing CAC but increasing LTV, ultimately you’ll see higher profits.
Increases email signups
Experimenting with the phrasing, timing, and placement of your call to action by asking website visitors to subscribe to a mailing list or register an account, you’ll be able to optimize your signup rate and increase the size of your mailing list.
Increases average order value (AOV)
Average order value (AOV) is another key performance indicator for ecommerce. Referring to the average value of each order placed, a high AOV is a sign of a healthy ecommerce strategy.
Reduces cart abandonment
If a visitor to your website adds an item to their basket, it means they’re at least considering purchasing it. To maximize the chances that they’ll complete the purchase, you can test different offers, layouts, and pop-ups to find out which one most reduces the number of abandoned carts your ecommerce site generates.
Things you can A/B test on ecommerce websites
In the field of ecommerce, a call to action (CTA) is a clear command or instruction directed toward your website visitors. Familiar examples include “buy now”, “add to basket”, and “enter your email for an exclusive discount”.
If you’re an ecommerce web designer, you need to be incorporating CTAs across your website, from landing pages to checkout pages.
But which CTAs are effective isn’t always clear. This is where A/B testing comes in. By trialing different CTAs, as well as their placement, you can figure out which ones lead to the most conversions.
Try to be original not just in the wording of your CTAs but in their delivery too. Pop-ups, banners, and carousels all offer opportunities to deliver a key CTA that could generate revenue. A/B testing gives you the opportunity to try new approaches. If they don’t work, you’ll soon know and can move on to the next CTA master plan.
Your landing page is the first thing people see when they arrive at your website. Thanks to the fact that the majority of ecommerce websites have been testing their landing pages regularly, we already know some of the techniques that work.
Landing pages should make it obvious what it is that you sell but shouldn’t be too heavy on text. For example, the landing page for the company I work for, Global App Testing, provides an overview of the services they offer but is not the place to start going into more complex topics like what is a smoke test, and other more in-depth testing related content found elsewhere on the site.
Landing pages are one area where very slight changes can make a big difference. So, it’s important to test everything from the CTAs you use to the amount of white space visible.
Video and images
Along with a chatbot for customer service, using videos and images is a sure way to help drive sales on ecommerce websites. Alongside following product photography best practices, A/B testing is essential to making your product images work for your website.
Many companies have had success using a combination of videos and images to show off their products. For example, it’s common for fashion retailers to compliment images of items with videos of them being worn, in order to give the customer a better idea of what it will look like in context.
You may not know it, but data from A/B testing has already transformed the way ecommerce retailers display their videos and images. For example, data collected by various analytics agencies has shown that vendors who remove the background (a street scene, for example) from their images experience up to a 34% improvement in conversion rates.
Different combinations of video and image, as well as variations and placements, can be tested.
Product descriptions are one aspect of ecommerce websites that is important to test because the customer base of every retailer will have different preferences.
For example, a data analytics provider selling integrated gaming solutions might find that concise but informative bullet points offer a higher conversion rate than lengthy paragraphs. Or companies that have more international customers might find that simplifying the language they use is the thing that helps to boost their numbers.
6 steps for successful ecommerce A/B testing
Having clearly defined goals is the foundation of a good ecommerce A/B testing strategy that allows you to design measurable, results-focused tests. Ensuring your tests are oriented towards specific goals is the testing equivalent of deploying performance marketing for your website.
If you’re wondering what is performance marketing? It’s a form of marketing that businesses use whereby they pay their marketers based on how well a given strategy yields a specific result - usually conversions.
We’ve covered some ground regarding conversions in this post but what do we really mean by this?
As well as converting website visitors into paying customers, conversion rates you can optimize for include visitors who add at least one item to their basket, the percentage of those who continue through to check-out, and how many of those ultimately make a purchase.
At each stage of this customer journey, you‘re both losing and winning a percentage of leads. When designing your tests, the first step is to anticipate a profitable customer journey. Ask yourself where the moments of truth are that decide whether a customer moves forward or not, and which elements of that stage in the journey can you modify and test? Planning things with a customer orientation strategy can certainly help you create actionable goals.
Once you’ve identified your goals, the next step is to make hypotheses. An A/B test is an example of statistical hypothesis testing. This means that a hypothesis is made about the relationship between two data sets and those data sets are then compared against each other to determine if there is a statistically significant relationship or not.
Hypotheses will be statements that can be tested - “more people will click on this button if it is red rather than green”, for example, or “fewer people will close their session at checkout if we remove the need to input an email address”.
For each hypothesis you’ll need to create the different versions you plan to test. You should design different versions for each and every element you wish to test.
There’s no point having one CTA on a red button and a differently-worded CTA on a green button as then you won’t know what the determining factor of people’s response is - the button colour or the wording of the CTA itself.
Run your A/B test
These days, most ecommerce web designers use dedicated tools to run their A/B tests. For a basic website, Google Analytics is the most popular solution and offers many of the features you’ll need to run simple A/B tests and analyse the data they provide.
For example, you’ve built a custom mobile shopping app, so you’ll likely enlist the help of a dedicated app testing service that runs more advanced automatic tests. If this sounds like your ecommerce business, you should research how to start automation testing from scratch.
The point of making hypotheses for your A/B tests is not to prove or disprove a given hypothesis for the test cases. Rather, it is to generate insights that can be used to determine whether or not there is a statistical likelihood that the hypothesis will be true or false for the majority of all cases.
Predictive statistical analysis is the science of predicting outcomes we don’t know using the information we do know. When it comes to A/B testing, it’s applied to make predictions about a whole population (total website traffic) based on the variance between your samples (the visitors who got shown different versions of a test case).
There’s no way of knowing with 100% accuracy how the next 50,000 people who visit your website will behave. What you can do is observe 1,000 people who visit your site and use statistical analysis with the help of automl applications to predict how the following 49,000 will behave.
You’ve set your goals, made your hypotheses, run your tests, and analysed whether the hypotheses are likely to be proven or disproven given the data gathered. The final stage is to use this insight to make the changes that your analysis predicts will have the optimal outcome when scaled up to the totality of your website traffic.
With an effective A/B testing regime installed, not only will you be able to make content decisions safe in the knowledge that your choices are backed by the data, but you’ll also be able to execute a full-scale website quality assurance program that ensures the optimal customer experience and customer retention. That’s because testing will flag up areas where visitors are leaving your site and highlight the design choices that are causing friction.
Now you know all about effective ecommerce A/B testing, all that’s left to do is implement your new knowledge in creating a business proposal for an eCommerce website or enhancing the one you already got.
With all the benefits discussed here up for grabs, what’s stopping you from applying the best testing methodologies and frameworks to your eCommerce website?