A/B testing in ecommerce: 10 things you really need to test
How do you know if everything is in the right place in your online store? Do you use the appropriate words when making an offer? Do you use the right colours to influence “properly” the subconscious of the users? Without testing, you do not know that: the best you can do is guessing. On the other hand, with A/B testing the relentless data will help you optimize – and we are going to help you now with testing.
This is what we are going to talk about:
- What are A/B tests?
- Where can you apply A/B testing?
- The process of testing
- What is worth testing?
- Visual elements
- Pop-up windows
- Urging solutions
- Payment solutions
- Forms, fields to be completed
- And everything else
Let’s start the story at the beginning: what are A/B tests?
As its name implies, A/B tests are trials, during which you try out two different variants of a given element. For example you use a red CTA (call to action) button on one of the pages and a blue one on the other. You place the search field at the top on one of the pages and to the right on the other.
Criteria of a good A/B test:
- Test only one element at a time – otherwise you will not know what the changes in results are due to.
- Run the two pages parallel and simultaneously – it may easily take an effect on your results if you run the tests with a difference of a few days. The test should run at least for a week.
- Use a significant sample: in case of public opinion surveys the minimal limit is 1000 people, this is the optimal for you, too.
- Distribute the visitors arriving at the two pages randomly: if you select based on segments, channels or any other methods, your results will be distorted.
It will be obvious after the testing which of the two solutions your customers like better: which encourage them to remain on the page, to register, to click through or even to purchase.
However, not only your results will be better. It also has a positive effect on the work ethics of the employees of your company that they do not have to rely solely on their intuition and the results are not requested from them while all they can do is guess: you can work with accurate data and precise analytics instead.
Such things may be inspiring for those in marketing and for the designers, because you do not let go of their hands while demanding results.
Where can you apply A/B tests? Practically everywhere
There is no element on your web-site, be it text, graphics, design or function, whose effectiveness you can be sure of without testing.
Usually by having it done by a professional. Even free trial version tools are available that can help you (for example Optimizely). However, if you work with a complex online store, you will need a developer, especially because this is the only way you can make sure that the criteria above will fully be realized. In the following, you can read a list of ideas about what kind of elements you can run A/B tests on. You can run several of these simultaneously, but you have to accept that you will not be able to analyse everything at the same time. Do not overburden yourself with data, try to achieve as good results as possible while collecting and utilizing information of a humanly processable quantity. The number of visitors may limit the testing: if you run 3-4 tests at the same time on a page, the visitors will be split into 6 to 8 parts, which can extremely elongate the duration of the test. Provided that you want to work with a significant sample. (Yes, you do.) First you have to be clear regarding your intentions: what would you like to achieve? It is not at all the same if you want the visitors to stay on the page for a longer time, or if it does not matter how long they stay provided that they execute a certain action. Once you know what you want, list the main elements of the relevant pages and prioritize them according to importance. You can monitor the analytics already at the definition of the priorities: if for example you notice that the majority of the visitors get to the CTA button on the page but they do not click on it, you will have to find out what may cause this. For the launch of the test you should have an already existing page, for example a product page. Select an element that you have doubts about: a CTA, a block of text, a form to be filled in. Make a new variant of it, try to make it more effective and you will have your A and B pages right away.
Process of testing: know what you want an prioritize your goals
Usually by having it done by a professional. Even free trial version tools are available that can help you (for example Optimizely). However, if you work with a complex online store, you will need a developer, especially because this is the only way you can make sure that the criteria above will fully be realized.
In the following, you can read a list of ideas about what kind of elements you can run A/B tests on. You can run several of these simultaneously, but you have to accept that you will not be able to analyse everything at the same time.
Do not overburden yourself with data, try to achieve as good results as possible while collecting and utilizing information of a humanly processable quantity.
The number of visitors may limit the testing: if you run 3-4 tests at the same time on a page, the visitors will be split into 6 to 8 parts, which can extremely elongate the duration of the test. Provided that you want to work with a significant sample. (Yes, you do.)
First you have to be clear regarding your intentions: what would you like to achieve?
It is not at all the same if you want the visitors to stay on the page for a longer time, or if it does not matter how long they stay provided that they execute a certain action.
Once you know what you want, list the main elements of the relevant pages and prioritize them according to importance.
You can monitor the analytics already at the definition of the priorities: if for example you notice that the majority of the visitors get to the CTA button on the page but they do not click on it, you will have to find out what may cause this.
For the launch of the test you should have an already existing page, for example a product page. Select an element that you have doubts about: a CTA, a block of text, a form to be filled in. Make a new variant of it, try to make it more effective and you will have your A and B pages right away.
What is worth testing?
1) Visual elements: make them as appealing as possible
Visual resources are basic elements of a product page. What is good about shopping in person is that you can touch the goods, look at them from every angle, you can examine them with all of your senses. In case of online shopping the salesman has to strive to deliver this experience as fully as possible. If you are selling a leather belt, use images or make videos that will make your visitor feel the scent of the real tanned leather even if only sitting in front of the monitor.
Of course the question is what kind of solutions work best. It is advisable to use as many and as high quality images as possible. This is done, and still you are not at all done with your possibilities, however, you cannot squeeze everything into one page. A fine balance should be achieved in the end.
You can use
- rotatable models,
- longer how-to videos,
- 10-15 second demos,
- or even sound files.
And you can test all these solutions, even according to content: it is not at all the same if you embed a demonstration video or a how-to video with tips in the page.
2) CTA ‒ What do your customers want to click on most?
In general, you would like the users to click on something. This is not as easy to achieve as you might think at first. What characteristics does a CTA have? Position, shape, colour, text… And each and every one of these can have a great effect on conversion.
If you place the CTA in the wrong place, less people will click on it. (According to the generally accepted rule, the more complex the offer, the lower it will be placed on the page.) However, it may also be fine if there are even two or three CTAs on the page, because each member of the target audience comes to a decision at different sections of the page.
If the CTA is too small, it will not be visible enough, if it is too big, it will be deterrent.
Colour is also an important factor:
According to the colour theory, some colours have an urging effect, some strengthen your sense of security, and so on. But it also may be that the CTA simply does not differ enough from its surroundings to be visible…
And you can fine-tune the text to the extreme: every single word has significance, they all have different effects on your audience.
3) Navigation ‒ drop-down, breadcrumb, categories…
You can apply a number of various navigation elements within a page – this way you can find out how your audience reacts to each of them.
It may be a question
- Whether you should put the categories on the right or on the left side.
- Whether you should use a drop-down menu or perhaps a floating solution.
- You may even test which categories you should position higher within a menu, which should be more in focus, because they can bring more clicks.
- If you use breadcrumb navigation, there are various available of those as well (path-based, hierarchy-based or attribute-based).
In this case you have to monitor how much time the users spend on the page, how many sub-pages they view, how they behave, in other words, what they click on within the page. If they click through more to your other pages in case of one of the solutions and less of them leave the page, you have found the better navigation.
4) Pop-up windows ‒ annoying but effective?
You can use the pop-up windows for many things:
- to request registration,
- to encourage to social sharing,
- to offer similar products or services,
- to prevent the users from leaving your page by offering an immediate discount.
However, it does make a difference what kind of it you use – a window that covers the whole page or one that appears discreetly in the lower corner?
When do you display the pop-up window? After a certain time or when the user makes some sort of action, or maybe when you feel passiveness?
5) Urging solutions ‒ coping with buyer hesitation
Try how a little urging influences your customers!
In our exhaustively detailed article we wrote about how you can do that gently: by indicating that a given offer is valid only for a certain time period (even by using a counter), by displaying the number of pieces or another way. See: 5 proven strategies to overcome buyer hesitation
Make variants of your already existing product pages where you place one of these elements – but make sure that your different pages do not contain different information. It may be annoying for the users if they view a page twice and at one time they see a limited time period for an offer and at the second time the counter is not there anymore.
Therefore, if you take a product page that has not included a counter previously, make two variants that show the limited offer in different ways (one of them for example by using an active counter and the other by simply indicating a deadline) and monitor whether the conversion increases compared to the previous data and also in which case by how much.
6) Payment solutions ‒ the simpler for the shopper, the better
In certain ecmmerce stores the customers have to fight their way through four or five pages until they can pay and order the product, in others, one or two pages are enough to be able to do that.
Too many pages can increase the number of cart abandoners, which is why it is worthwhile for you trying whether you can shorten the process somewhere, and testing the solutions of more and also of fewer pages.
This may be an especially important factor if a lot of your visitors arrive from mobile devices: they are usually more impatient, so having to carry out an action through several pages significantly spoils their user experience. In their case you can try for example a solution where they can go through the whole procedure on a single page.
7) Forms & fields to be completed
According to certain studies, your conversion rate decreases by approximately 10% with each additional field to be completed.
Users do not like when you ask for too much data – on the other hand, you can make much use of that data during the segmentation later in email marketing.
So test how much data you can request “with impunity” – can you ask for example position, company name, city or zip code beside the name and email address?
It is even more true in case of sensitive information that people are less willing to provide them – also test whether you are better off with mandatory fields (this may result in somewhat less leads but a list that can be more precisely segmented) or with optional ones (with as many email addresses as possible but with incomplete data).
By requesting registration for being able to make a purchase, you usually increase significantly the number of cart abandoners and bouncers. In certain cases however, this may be a loss that you will profit more from in the end.
It is possible that you will have less customers like that, but you will also be able to target them again more effectively later: you will have more data, you will be able to contact them through multiple channels, you can rely on their previous behaviour more securely.
And those who have already purchased from you, can always be contacted and converted again more cost-effectively than new customers. Therefore you should test mandatory and optional registration and calculate whether it is worth more in the long run sacrificing a few newcomers in exchange for the customers that you will be able to reactivate more easily later.
Test what sorts of texts work well on your product pages! The long ones that work with storytelling, or the shorter and more to-the-point copy brings more conversions?
What style do you write in?
If you offer business services to large companies, perhaps the official tone will bring you more customers. On the other hand, if you are selling skateboards, you will most probably be better off with a cooler style.
You may try the negative and positive tones as well. There is a huge difference between “Buy it now!” and “Add it to cart or you will miss it!” Some customer segments may react negatively on such urging, while in other groups such scarcity can significantly increase conversions.
What kind of headlines, page titles do you use? It does matter whether you attract attention with creative or informative solutions.
It does make a difference what kind of offers you make to the users. We have already written about holding back information being responsible for an important part of cart abandonments, just like many times the non-indication of delivery costs.
It is possible that many people will add the product to cart if they do not see that they will also have to pay for delivery, but many of them will leave it in the cart unpaid as well, because they will feel you betrayed them if you make them realize the “real” or total price only at payment.
Consequently, you can indicate the shipping cost on the product page, moreover, we even suggest you do so – however, you might as well try offering free delivery under given conditions, for example if customers purchase products for above a certain amount, or order a given number of pieces at once.
You can increase the average value of your orders by that, given which free delivery will already be worthwhile for you, and at the same time you do not hide the costs either – but it is wort a thought whether you will lose more customers by that and it may be more beneficial to direct them towards a larger order.
+1: And everything else
There is practically no element whose operation you could not test, however, as we have already mentioned at the beginning of the article, you cannot test everything simultaneously.
Find out which points are critical based on the data:
- on which pages the most customers bounce back?
- for which products are there the least conversions?
- where does activity fall or soar spectacularly following a given modification?
It is recommended that you try your own page yourself as well: examine where you look, where you click, which are the points where you yourself would drop the purchase.
A/B testing can be a real secret weapon, studies clearly show that you can improve the conversion rate of a page by a percentage of at least two digits with a single tiny modification.
To be able to realize that, you will have to carry out the testing smartly though – probably it will be best for you to start with the tips described in this article.
With only a minor change on your product page you can considerably improve conversion. Click To Tweet
If you consider implementing an A/B testing solution for Magento stores, the extension “A/B Testing” developed by Amasty seems a good option. You can run different split tests and multivariate tests in terms of design or different attributes, can set test duration and can have statistical reports about the outcome of the tests.
Thanks to its world-class features and robustness, it is probably the most popular Magento extension for A/B testing.
- Run numerous a/b tests simultaneously
- Test particular attributes for a group of products at a time
- Test product page design
- Get informative stats reports
- Manage test duration
- Display test variants to customers according to conversion rates
We hope that the practical examples in the article put in the mood for A/B testing even those who had previously considered them to be unnecessary.
You can be sure that the changes you make based on these results will be felt in increased conversion rates.
If you have any questions, do not hesitate to ask away in the comment section or contact us, we are glad to help.
Digital Marketing Manager
Gábor started working for AionHill as a Magento eCommerce project manager. After 6 years, he joined the restructured Marketing Team. His main responsibilities are content strategy and communications partnership management. Gábor has two kids, likes cooking traditional Hungarian meals and playing strategy and online role playing games.