What are A/B tests?
A/B testing is a technique used mainly in marketing, but also in graphic design, IT and many other areas. It involves comparing two or more variants under the closest possible conditions and verifying which one works better.
A typical A/B test involves displaying the element being tested – for example, a button linking a user to a newsletter sign-up form – to two groups of site visitors at the same time. Half of them see variant A, the other half are shown variant B. After a predetermined period of time (usually the tests last from a few days to a few weeks), the conversion results are compared and the winning variant is implemented permanently. An extremely important rule of conducting A/B tests is that no more than one element should be tested at a time – this way the results will be authoritative and will allow to see how significant the implemented change actually turned out to be.
The goal of A/B testing is usually the same: to achieve higher conversions, but not all elements of the store affect conversions directly. This does not mean, however, that you should abandon testing them. There are many tools that help verify the effectiveness of a given element in other ways than simply checking the conversion rate value. Specialists often use, for example, heat maps, which indicate how users use a website – where they stop their cursor and which elements of the site they focus on the most.
Which elements of the store can be tested?
An online store is not just the website itself, where customers select products, add them to their shopping carts and place orders. It’s also newsletters, emails with promotions, landing pages and a whole range of social media where we can share standard offers and special actions with regular and potential customers.
A/B testing can be successfully applied not only to the website, but also to off-site elements.
What is worth testing on the store’s website?
- Headlines – what the customer sees at the very beginning of a visit to a website is the headline. Its role can hardly be overestimated – a properly written headline can entice a customer to buy, while a careless few words can make him look for another store all at once.
- Product names – it happens that the name of the product is misleading, so that it does not find the right number of buyers. What’s more, when selling electronic devices, for example, you have to reckon with the fact that only the most relevant information should be included in the product name – all of it can’t fit anyway.
- product descriptions – testing several thousand product descriptions is, of course, Sisyphean work, which may not at all bring noticeable results, but if you notice that a product in your store sells poorly, while elsewhere it is a source of great revenue, it is worth looking at the situation a little closer and – perhaps – improving the product description. Learn how to use ChatGPT to create product descriptions.
- Other content – it’s worth taking a closer look at all content elements. Notifications, pop-ups, banners – any of these could potentially work better and result in better results.
- CTA buttons – Call to Action is one of the most important places on a website. It is the clicking of this link or button that produces the desired conversions. There are quite a few buttons in a store: from the most iconic “Add to cart” through “Place order” to those less directly affecting sales, but equally important, such as “Add to wish list.” It’s not just the content of the button that’s worth testing – it can be just as important to change its color, shape or placement.
- The visual layer of the site – the effectiveness of the site is also affected by its visual layer. It’s not just the images on the homepage or product photos, but also the color scheme of the site, the fonts used or the placement of elements.
- Navigation – the menu of an online store can be very helpful, or, on the contrary, cause users a lot of trouble. It is worthwhile to regularly check how customers use the menu and take some time to test different variants that can make it easier to navigate the store.
What elements beyond the store page can be tested?
It’s not only on the store site itself that we can conduct A/B testing. In addition to the store’s website, it’s worth testing content in newsletters, the design and content of ads, and even social media posts – some platforms, such as Facebook, even have built-in tools to conduct A/B tests.
How do you prepare to run A/B tests?
Before you decide to conduct the first tests, it is worth making some preparations. Determine what you want to start testing with. The goal of testing is to increase the conversion rate – but remember that sales are only one type of conversion. Not every element directly influences sales – some are designed to increase brand trust, while others lead to micro-conversions, such as newsletter sign-ups (which for the store means acquiring leads and expanding the potential customer base).
Once you have a specific test goal – for example, to increase the number of new customer registrations – think about which elements of the store affect this conversion. For example: on the website, it will certainly be a button or link that encourages registration, as well as a window that appears after an order is placed. Off-page, it could be an ad announcing a discount for new customers, or a mailing sent to leads, with a code entitling them to receive free shipping upon registration. Take note of which of these elements are not performing as expected and start with the ones that are doing the worst.
Once you’ve completed the list of items to improve, rank them from those that can be tested with the lowest cost and effort to those that are the most labor-intensive.
How to conduct A/B testing in your store?
Take the first item in the list and plan actions for it. If it’s a button whose content you want to shorten, offer some variations.
Example
Currently, the button says “Create a new user account.” You can suggest shorter and more specific content, such as “Create an account” or “Register”, but this is not the only way. Perhaps you’d prefer to test a button with more casual content? Try: “Become a customer” or “Join us”!
A/B testing gives you a lot of freedom and allows you to test even the most unusual variants of content and page design. Access to ongoing tests is usually unlimited, so if you notice that the new version has not won the approval of users, and the initial results already show a large discrepancy, you can always stop the test.
Sometimes the term “A/B testing” is used instead of “A/B/X testing” – that’s because you can compare more variants during one test. Unsure which button color works best with your site’s layout while standing out from the rest? Run the test with several variants, dividing the target group equally. Remember, however, that the more variants, the larger the group should be. If the page is viewed by 1,000 people, two versions of the button will be seen by about 500 people. However, if you prepare seven variants, each of them will see only about 140 visitors, which will reduce the credibility of the test.
The next step is already running the test. Fortunately, you don’t have to bite into the site’s code to do this – there are simple tools that allow you to run tests, manage them, and display easy-to-interpret results.
Once the test is complete, all you have to do is implement the change (or leave the older variant, if that’s the one that scored better). This is a relatively easy process, but it is not without risk.
The pitfalls of A/B testing
The general rule of thumb, often repeated by marketing or e-commerce specialists, is that you should test virtually everything, and as often as possible. This is generally good advice, but only under the right conditions. You can try to apply it when you run a large online store and have a team that has the capacity to improve various aspects of the store.
Otherwise, it is easy to fall into the trap of testing that does not translate into real changes for the store. This is especially true for subpages or products that are not frequently visited – A/B testing should be done first on the most trafficked ones and those that have the highest impact on conversions.
Organizing A/B tests, including time-consuming preparation and analysis, only makes sense if the changes made will bring noticeable results behind them. Of course, it is not always easy to predict the results of a test, but if you spend a few hours testing a sub-site that records 200 unique visits per month, it will more than likely be time wasted.
Summary
A/B testing can help your online store. It’s a powerful tool that has great potential to improve conversions in key areas of your e-commerce. The most important thing in getting satisfactory results, however, is to set a goal, make a list of elements that testing can have a significant impact on your store with little effort needed to make changes, and then stick to the findings made.
