Conducting A/B tests is one of the fastest and most effective ways to learn about the preferences of your website visitors. Robin Pratap, founder of the German online shop asanayoga.de, shows exactly which measures increased his sales.
During an A/B test, you test the current webpage or an online shop (original) against a variation with diversions in aspects like text, images or layout. The aim is to find out which version is more beneficial for your shop or business. This method is used mostly used with general conversion improvement or when redesigning the portal.
Assumptions or Knowledge: What do my Costumers Want?
The big advantage is that conversion optimization with A/B tests completely fueled by data. Success is measured with goals like product sales, newsletter subscriptions or downloads. That way you can save empty phrases like “I think that image appeals more to our target group than the other one”. You can make comparisons using testing tools within minutes. After that, the validation is just a matter of time and statistical significance.
Especially in the highly competitive world of eCommerce, it is important to compel visitors. The traffic was either dearly bought through expensive AdWords measures or generated through long-term SEO or social media efforts and it would be a shame to waste that. In the following case study from the yoga industry we want to show you how you can increase revenue in an online shop environment by 50 % simply by optimizing a landing page.
A/B Test Tools for WordPress
We are devoted WordPress WooCommerce und WooCommerce German Market users ourselves. For A/B tests there are also various services that can be easily integrated with a plugin or a line of code in WordPress. Some providers are
- Convert with its WordPress integration
- Optimizely with the Optimizely plugin
- The Visual Website Optimizer (See this blog post on integration into WordPress)
Before we started A/B tests, we tested some providers and quickly decided to go with the Convert solution. What convinced us was the setup, the simple Google Analytics integration (including aims) and the usability of the tool.
Having a measure – how do I get relevant data?
Every A/B test starts with a hypothesis, something like “With an emotional hero image on our landing page, costumers are more inclined to buy our product.” Here are some ideas for the first tests:
- Different images and structure for important landing pages (compared to other shop or portal pages)
- Comparing short and detailed product descriptions.
- Color of buttons and elements – the classic among tests is changing the color of the „Add to Cart“ button.
- Adding trust seals to the shopping cart (like 30-days money back guarantee, Safe Shopping, guaranteed satisfaction, etc.)
There are a huge number of test options with lots of potential. For this, every opportunity should be taken to achieve sustainable and effective conversion optimization.
Once you have decided on a hypothesis, you can start setting up the test. Usually, all A/B testing tools give you the following options for comparing two versions:
- Simple A/B test – Two or more versions are compared to the original. The variations are created and hosted directly in the A/B test tool (simple editor).
- Split URL test – For more complex experiments (like major layout changes) you can create a new page for the test in WordPress and test that URL. (IMPORTANT: Mind the Dos and Don’ts at the end of the article for this option.)
- Multi-variant test – Testing different combinations on one page to better understand user behavior.
In all A/B tests, the traffic is distributed in equal parts to the original and the variation(s). It is vital to consider the Confidence Level. It tells you, when statistical significance is reached and valid data about the improvement or deterioration is available. If the Confidence Level is at 95 per cent or higher, your data is most certainly valid.
Practical example: 50 Percent Increase in Conversion in one Month
We are a yoga shop running the eCommerce store asanayoga.de, focusing on high-quality and sustainable yoga products. One of the most important yoga products is the yoga mat, for which we created a special landing page. Below you can see a comparison between the original and the variation:
For this test, we direct costumers searching for „yoga mats“ directly to the landing pages. Over the course of one month, visitors were directed to the original (left) or the variation (right). Apart from the landing page, all other pages are identical (product pages, shopping cart, etc.).
The text content of the variation is mostly identical to the original. The biggest change is in the structure and choice of images. The original mostly describes comprehensible elements and offers a comparison of products. Concerning the variation, it was important to us to choose a more emotional communication reflecting in headlines, structure and images.
Without further ado, here is the result:
In the main aim, Orders, the variation exceeded the original by 53.11 percent. This also means 50 percent more sales in this case. The conversion of 1,68 percent has increased to 2.71 percent – a great result, achieved only by a different landing page.
This diagram clearly shows that the variation outperformed the original at any time.
In our experience, it is generally recommendable to also measure other aims (apart from sales) with the A/B test. We have picked these additional goals:
- Orders (53 percent increase in the displayed case)
- Lower bounce rate (+ 5 percent)
- Higher activity (+ 15 percent)
- Shopping cart visits (+ 57 percent)
- Visitors at checkout (+ 34 percent)
- More than 5 minutes (+ 38 percent)
This showcases the non-proportional of visitors at checkout (“only“ 34 percent more) compared to a 53 percent increase of orders. This already leads us to the next hypothesis and test: Is there room for improvement at checkout?
Dos and Don’ts – Advice From Real Life
- Split URL: Watch your SEO ranking! You should set the variation to “no index”, so it is not considered duplicate content by Google. Also, the variation should by all means point to the original with a canonical. We use the WordPress SEO Plugin by Yoast, which allows these settings.
- Small vs. Big Picture: Don’t just optimize one page. In most cases, conversion is a series of user interactions. That means the connection between landing page, category page, product page, upsell, cross sell, shopping cart and checkout should always be considered in the big picture.
- Time will tell: By the time a test is successful, the conversion funnel may have already changed and old hypothesis may have to be questioned with new ones.
- Epic fail: It might happen that even after a long time there is not much difference between the original and the conversion. That’s a shame, but that’s life :). New game, new luck!
- Statistical significance: Every test should reach a confidence level of at least 90, ideally 95 percent. With that, your winning variation can go live with good conscience.
You can ask questions about A/B tests and conversion optimization in the comments. I’m also happy to hear your experience with A/B tests.
Role Based Prices for WooCommerce is available again: Different prices for different customer groupsby Michael Firnkes
You want different prices for different customers and users of your online shop? E.g. for subscribers or wholesaler? Or do you need bulk prices? A lot of o ...Read more
WooCommerce Weekly Review #7: Woo vs Magento, shop security and customer confidenceby Michael Firnkes
What is a WooCommerce shop in practice? And how do you protect it against external attacks? This and more in our Weekly Review. WooCommerce vs Magento: ...Read more
WooCommerce Weekly Review #6: Translations, WordCamp Switzerland & avoiding legal warningsby Michael Firnkes
WooCommerce was a hot topic at WordCamp in Switzerland. Progress is being made with the system translations and there are lots of useful new extensions. Th ...Read more
Inpsyde is a WooExperts Gold Partner of WooCommerceby Michael Firnkes
A few weeks ago, Inpsyde GmbH with MarketPress became a certified WooExperts Gold Partner. This is confirmation from WooThemes of our agency’s know-h ...Read more