Running a webshop means making countless decisions that can directly influence how your customers behave. By now it's widely accepted that it matters what product image greets a visitor, how prominent the add-to-cart button is, or in what form you communicate free shipping and price discounts. Most merchants rely on their own experience or on solutions they've seen at competitors — yet what works in one webshop won't necessarily deliver results with a different customer base.
A/B testing helps you stop changing your webshop based on gut feeling alone and start seeing how customers actually react to each solution. You can try out two different versions of the same page, for example, and the results can reveal which one works better for sales. You don't necessarily need a major overhaul for this. Often, modifying a single element is enough to understand your customers' behaviour better and find out what's worth changing to win more orders.
This method can be useful for webshops that already receive visitors regularly but want to squeeze more revenue out of their existing traffic. In this situation it may seem obvious to spend more money on advertising, yet it's worth taking a closer look at how the webshop itself works first. You may be getting enough interested visitors — just too few of them make it all the way to placing an order. If you can find out where and why customers get stuck, your sales results can improve even without raising your ad budget.
How does an A/B test work in a webshop?
The essence of A/B testing is trying out two versions of the same page or feature — at the same time. Visitors are randomly split into two groups: one sees the original “A” version, the other the modified “B” version. This lets you compare which solution delivers better results.
Let's take a simple example. A product page currently has an “Add to cart” label on the purchase button, but you're curious whether more people would buy if the text read “Add it to my cart”. Instead of changing the button for everyone, you show half of your visitors the old version and the other half the new one.
During the test you can see which group adds more products to the cart, and how many ultimately complete an order. Of course, you shouldn't draw conclusions from just a handful of purchases. You need a sufficient amount of data and statistical evaluation to judge whether the change can really be behind the difference. It's important to understand that an A/B test is not a simple before-and-after comparison. If the original version runs one week and the new one the next, the result can easily be skewed by ads that changed in the meantime, a different visitor mix, or even paydays. That's exactly why the two versions should be tested simultaneously, on randomly assigned, comparable groups of visitors.
What should you test first? Not necessarily the colour of the buttons
When A/B testing comes up, changing button colours is often mentioned as a simple, easy-to-try solution. In a webshop, however, that's usually not the most important question. You'll get much further if you first look at where customers get stuck and at which stage of the purchase process most of them turn back.
If, for example, many people view your products but hardly anything ends up in the cart, it may be worth experimenting with product photos, descriptions, or the way offers are displayed. The situation is different if visitors do fill their carts but never finish the order. In that case you need to find out what's holding them back; perhaps the shipping cost is too high, they can't find their preferred payment method, or the ordering process simply has too many steps.
Sellvio's traffic and sales statistics, cart value data and analytics integrations can help you figure out where to start investigating. That way you're not changing things in your webshop at random — you deal first with the problems that your available data is actually pointing to.
On the product page, a single piece of information can decide the purchase
When someone is browsing a product page, chances are they're already past the initial research phase. They roughly know what they're looking for, but they haven't necessarily decided to buy. At this point it can matter a great deal what images they encounter, how easily they find the details that are important to them, and whether they get answers to the questions that still make them hesitate. A/B testing is exactly the way to examine these seemingly small but often decisive differences.
Take product photos, for example; in a clothing webshop it can help a lot if the customer sees the garment on a model as well, while for technical products close-up shots of the details often provide more useful information. With product descriptions, the longer text doesn't necessarily work better either. It may contain every necessary detail — it's just hard to find the point. In an alternative version, the most important properties and practical benefits could come first. Whether that actually results in more purchases is something you should decide by testing.
With Sellvio's product management and content editing features, the individual versions are relatively easy to create, and the AI-based content tools can help here too. That said, it has to be stated clearly: creating two different product pages is not yet an A/B test; you also need to make sure the versions are shown to properly separated, comparable groups of visitors.
Whether a customer finally places their order often comes down to perfectly ordinary information: when will the product arrive, is it in stock, how long is the warranty, and what happens if it doesn't meet their expectations? It's worth answering these questions clearly, right on the product page. It can easily turn out that a more thoughtful placement of shipping, stock and warranty information improves conversion more than a completely redesigned product page.
Category pages: do customers easily find what they're looking for?
It's often forgotten, but in a webshop offering hundreds or thousands of products, the quality of navigation can have a significant impact on sales too; visitors often arrive without a specific idea, so they need search, filtering and comparison to find the right product. It's worth testing the default order of products, because it matters whether popular products, new arrivals or the bestseller list is put front and centre. A similarly important question is which filtering options should be visible immediately and which are better placed in a collapsible menu.
Sellvio's category, product variant and filter management features support building a transparent product catalogue, while the customisable design options help you tailor the presentation to your customers' needs.
And don't forget: examining mobile usability is especially important, since a category page that works well on a desktop computer can easily become overcrowded on a phone — so the simplicity of navigation, the accessibility of filters and the information content of product cards can all be worth A/B testing.
Shipping conditions and checkout – where purchase intent is easily lost
We've referred several times to the Baymard Institute's aggregated cart abandonment research, according to which nearly 70 percent of online carts never end in a transaction. This figure, built on the results of 50 studies, shows just how much the purchase process matters. Although many people are merely browsing or comparing prices, unexpected shipping costs, missing payment methods and a complicated checkout process can genuinely hold customers back.
It's worth testing whether more customers make it to the order if they can already see on the product page how much shipping costs or when they can expect the parcel. It's also worth testing whether the purchase rate improves when parcel locker pickup or card payment options are more prominently displayed.
Sellvio can be connected to a range of payment and shipping services, and Sellvio Pay and Sellvio Express make managing them easier too, so merchants have more room to decide which payment and delivery options they offer their customers.
The checkout process itself, however, is not something to change carelessly — a badly executed test can easily cost you orders. It's better to first examine whether it matters where and in what form shipping and payment information appears. Only touch the payment flow itself once you've thoroughly verified the modifications beforehand.
Promotions and coupons: the bigger discount doesn't always win
A well-crafted promotion can give your webshop's traffic a serious boost, but the number of orders alone won't tell you how worthwhile the discount actually was. It can easily happen, for example, that a general 10 percent price reduction brings in more customers than a HUF 3,000 coupon that can only be redeemed above a certain cart value. Yet the latter may be the better deal for the merchant if it makes customers put more products in their carts.
Sellvio's coupon manager lets you set up several types of discount. You can offer a percentage reduction or a fixed amount, define a minimum cart value, or even restrict coupon use to specific products, categories or manufacturers. Promotions can also be scheduled, and volume discounts can encourage purchases too — so you don't have to use the same solution for every promotion, but can try out which one brings the best result.
With an A/B test you can also see which promotional offer customers respond to better. It may be worth comparing different free shipping thresholds, coupons or discounts, for example. When evaluating the results, though, it's not enough to look at how many orders came in or how much revenue was generated — you also have to factor in the cost of the discounts and the profit left on the orders. It can easily turn out that the promotion driving the most traffic ultimately earns you less. A well-designed A/B test is exactly what can shine a light on differences like these.
What should you measure? Conversion rate doesn't tell the whole story
In A/B tests we most often watch the conversion rate. It shows how many of your visitors make it to the desired action. In a webshop, a purchase usually counts as the conversion, but you can just as well measure how many people add a product to the cart, start the payment process, or complete some other action that matters to you.
Let's look at a simple example. If 10,000 visitors produce 200 orders, you have a 2 percent conversion rate. Suppose that after a modification, the same number of visitors yields 220 purchases. That's a 2.2 percent conversion, i.e. a 10 percent relative improvement. It looks encouraging at first, but you can't yet state with confidence that the change really worked — with a sample this size, pure chance can produce a similar difference. That's exactly why you need to examine the results from a statistical standpoint too.
Beyond the conversion rate, other numbers are worth watching as well. Average cart value says a lot, as does revenue per visitor — and, if the right data is available, the profit earned on orders. After all, it's no use having more people buy after a change if they spend less at the same time and the webshop ends up worse off overall.
A good starting point for measurement is Google Analytics 4, whose e-commerce events let you track the individual steps of the purchase journey. The view_item event records product views, add_to_cart records items added to the cart, begin_checkout the start of the checkout process, and purchase the completed purchases. Of course, all of this only gives a reliable picture if the events are set up and verified properly; simply adding the Analytics tracking code to your webshop doesn't guarantee that you're seeing all the data you need.
How many visitors do you need for a reliable A/B test?
One of the pitfalls of A/B testing is wanting to see results too soon. After a few days, one version may already look better, but that's far from a guarantee it would bring more purchases in the long run. This is especially true when you're only talking about a few dozen orders and trying to detect a relatively small difference.
Let's look at an example. Take a webshop where, on average, two out of a hundred visitors place an order — a 2 percent conversion rate. Suppose you want to raise that to 2.2 percent with a change. The difference doesn't seem big at first, but that's precisely why, statistically, it takes a lot of data. With a conventional two-sided test, a 5 percent significance level and 80 percent statistical power, you need roughly 80,000 visitors per variant. In total, then, about 160,000 visits are required to detect a difference of this size with adequate reliability. For a smaller webshop, that can mean testing for several months.
It's therefore worth considering what you test in the first place. If you have few visitors, there's not much point in running a separate A/B test on every tiny modification. It makes more sense to concentrate on the changes that can bring genuinely noticeable improvement. And finding the most important problems doesn't necessarily require a statistical test: customer feedback or a usability study can reveal a great deal too.
How can you run an A/B test in a Sellvio webshop?
Sellvio offers several options worth taking advantage of if you want to improve your webshop's conversion rate. You can modify the design, rework the content of product pages, try out different coupons or promotions, and the sales statistics give you a picture of what your changes achieved. These are a good starting point for preparing A/B tests.
To measure visitor behaviour, both Google Analytics and Google Tag Manager can be connected to the Sellvio system. With the right configuration you can track how customers use the webshop, which elements they interact with, and where they abandon the purchase. You'll also find help for setting up GA4 and GTM event tracking in the Sellvio knowledge base. If you want to understand visitor behaviour in even more detail, you can use tools such as Hotjar, Smartlook or Inspectlet. These can reveal, for example, on which page customers start to hesitate, or what usability problems make ordering difficult. One thing to keep in mind, though: measuring visitor behaviour and running an A/B test are not the same thing. Just because Google Analytics is collecting data, it won't automatically split your visitors into two groups and show each the right version. For that you need a dedicated experimentation tool or custom development that handles the random assignment of variants, the identification of test groups and the statistical evaluation of the results.
Where Sellvio helps most is in preparing and implementing the changes themselves. Since product content, design and numerous sales settings are managed in one place, building the variants you want to test can be simpler. And once the measurement results show which version works better, rolling it out doesn't necessarily require any serious development work.
A/B testing isn't a one-off task — it's a learning process
You can learn a lot from an A/B test even when the new version ultimately doesn't outperform the old one. In fact, discovering that a promising-looking modification would actually hurt sales is a useful result in itself. It lets you discard an idea before investing serious time or money in rolling it out — and all the while you get to know your own customers' habits better and better. This matters because what worked for another webshop may not work the same way for you at all.
Sellvio's content management features, analytics integrations and business reports can help ensure your development decisions aren't based purely on gut feeling. Of course, the goal isn't to constantly change something in the webshop. It's worth far more to understand which modifications make buying easier and which bring genuinely measurable improvement in sales.
The success of a well-run webshop doesn't necessarily come from a striking appearance or spectacular features. It matters at least as much how well you understand your customers' needs and what you do with what you learn from their behaviour. If you build your improvements on these insights, you can make your webshop more successful in the long run too.