A/B Testing (also called split testing or bucket testing) is a technique used to determine if a specific change made on a webpage will influence conversions in a positive or negative way. It is an important part of the CRO (Conversion Rate Optimization) process and it is implemented through the creation of a variation page (page B) whose results are compared with the ones of the original version (page A). The one that performs better in terms of conversions wins the challenge.
A/B testing and split URL testing are usually interchangeably used, as there are no major differences between them. A point of difference, however, may be that A/B testing takes place on the same URL, while split testing uses traffic segmentation to randomly distribute visitors between two different URLs of the same page.
In A/B testing, a control version is compared to a second one, called the variation which features minor, contextual changes of the main version. In split testing, the second version is completely different from the control version. The main purpose is to find the winning variation, the one version that will improve conversions and help you most with your online marketing efforts.
The importance of A/B testing
The purpose of A/B testing is to conduct experiments leading to improvements in the user experience, while also collecting valuable data on the results. It leads to a deeper understanding of users’ behavior and which on-site elements impact it most. A/B testing also allows us to better reach our conversion goals, as long as it’s carried out systematically, in a constant attempt to improve any given experience on the website.
The A/B testing marketing effects are also significant: they can positively impact your SEO efforts, as well as your copywriting techniques. A/B testing can also be used for newsletters or sponsored ads. Marketing and A/B testing have a lot in common, actually: both of them are trial and error processes that build on the existing knowledge on consumer behavior and help expand it. And since we brought SEO into the discussion, did you know that the changes you make can improve the loading speed of your website or your landing page? A delay of only 1 second in the load time means a 7% reduction in conversions.
Yet there are numerous other advantages to A/B testing, that go way beyond user experience:
- Get more conversion by investing less. You don’t need to spend more on advertising, you need to focus on obtaining more qualified leads. A/B testing allows you to get the most out of your existing traffic, as even minor changes can have a great impact on your conversion rate.
- Reduce bounce rates on your website. The longer a user stays on your website, the higher the chances that they will end up performing the action you want them to. Whether you want them to subscribe to your mailing list, to give you contact information in exchange for promotional discounts, to book a demo or to buy a product, the principle still stands. Smaller bounce rates improve your chances of making a good impression on users and allow you to gain momentum and win their trust.
- Only make minor modifications that you’re sure of. A/B testing allows you to make low-risk changes, like playing with the button sizes or the button color, instead of redesigning the whole page or experience. This, in turn, reduces the risk of compromising your current conversion rate. Because let’s face it, there’s always a chance things would go wrong.
- Get a sense of how you can profitably redesign your website. A/B testing can offer you useful insights regarding which color scheme, copy style, fonts or designs work best. They can be used as a solid starting point if you ever consider redesigning your website. Whenever you improve your website’s UX, you take a healthy step towards boosting the confidence level of your users.
A/B testing benefits
Now that we know what is A/B testing, it’s time to talk about the benefits of this method.
- Reasonable costs and benefits. The test in itself is not expensive, you just need a good platform for A/B testing (some of them don’t even require programming skills) and the analytics tools that are already in place, like Google Analytics. You only need to take the time and define the hypothesis and the variables to be tested, then, if you carry out the experiment in-house, your UX specialist will perform the test. If not, you can always outsource the process to a specialized company and this way it will become even more cost-effective, by eliminating the risk for something to go wrong, due to lack of experience or poorly designed hypotheses and variables.
- Reach your goals faster. This is the main purpose and the final outcome of A/B tests and CRO in general. By performing these tests, you can gradually modify more elements on your website and your landing pages, by making them more appealing and relevant to users. This, in turn, will lead to more conversions (subscribing to a newsletter, visiting a landing page, clicking on a CTA, purchasing a product).
- Increasing the efficacy of sales funnels through funnel testing. Every business and every eCommerce website should be based on the principles of the sales funnel. A sales funnel is a way to guide your users throughout their entire customer journey, from the awareness and interest stage, towards consideration, intent, evaluation and then, eventually, purchase. When users arrive on your website, they are merely visitors/prospects. Through funnel testing, you can emphasize the elements which have the potential of converting them into leads, MQLs (leads that are the most likely to turn into customers), SQLs (highly engaged and interested leads, one step away from becoming customers) and, ultimately, into customers.
- Healthier websites and landing pages. Good marketing is based on trial and error. Only through such experiments can we determine which elements perform better and have the highest potential of positively impacting consumers’ behavior. By carrying out such tests, you can improve the quality of your website, as well as your overall strategy, because you get to focus on what matters most to the people you are trying to reach.
- Eliminate pain points from your users’ journeys. In order for users to perform the action they are supposed to or to accomplish the goal for which they came on your website, you need to make the experience as simple and intuitive as it can be. A more user-friendly design, clear product descriptions and a visible CTA button can take you a long way.
- Maximize the results of your ad campaigns. Split testing allows you to test your current campaigns with your real audience and help you gather precious data to back up your future decisions. It’s easier to meet your goals when you focus on your customers and make improvements that are relevant to them.
A/B tests or multivariate tests?
A/B tests are quantitative tests that work with a single variable. Most of the time, you only need to compare two versions in order to determine the changes that have to be made. However, this variable doesn’t have to have only two versions. The change can be implemented on a series of pages: A/B/C/D/E/F/G/H/I/J and so on. For example, you can test up to 10 versions of a call to action, by splitting the traffic to 5 versions of the same page, each one with a different call to action.
In contrast to A/B tests, Multivariate tests use multiple variables to find out which combination of variables converts more. For example, on one variation you can change the title, the call to action button and the price at the same time. Testing multiple variants can be useful when you don’t know exactly what calls for improvement and you’re gradually trying out different versions of different elements. Yet this kind of test is rather hard to validate and is not recommended for websites that don’t bring in high volumes of traffic.
A/B tests are recommended in the situation where you can point exactly to the element that could be the key to increasing conversions.
How to conduct A/B testing
We will provide you with a useful guide to A/B testing right away but there are a few good practices we need to discuss before you get started.
There are a few basic principles guiding the A/B testing methodology. First of all, you need to know what metrics you need to improve. Conversion metrics differ from one industry to another. For a good start, you may want to check the industry’s conversion rate benchmarks in order to discover how well your website is performing, compared to the standard rates within your field of activity.
You also need to have that conversion funnel in place. It’s that one tiny hack that will help you make the most out of your leads by identifying the specific steps of the buyer’s journey attached to each consumer avatar and making changes according to them. People come to you from different places and they want different things as well. Maybe you want to gain more awareness, so you should focus on the awareness and interest stages of the funnel. Or maybe you need to do something with that cart abandonment rate, so the optimizing efforts should be focused towards the intent and purchasing phase.
Then, it’s useful to research the most common reasons that affect your website’s performance. This usually happens in two phases: the quantitative and the qualitative phase. During the quantitative phase, you will look into Google Analytics (or whatever analytics tool that you are using) in order to highlight the areas on your website where traffic and conversions are lost.
The qualitative phase involves researching the patterns of visitors’ behavior. What do users usually look for when they come to you? How easy is it for them to find that? What makes them spend more time on a website, thus decreasing the bounce rate? What guides their buyer journey and how can you improve on-site elements in order for them to better meet the requirements of each stage of this journey? These are just a few questions you may need to answer before coming up with an A/B testing plan and setting up the whole process. The answers will come from user surveys, customer surveys, usability testing or whatever qualitative methods you find fit for your specific case.
How does univariate testing and analysis work?
Though apparently a simple technique, A/B Testing is a process that includes an in-depth data analysis and gives essential insights into the way visitors perceive and interact with the site. Results will appear only if the test is implemented correctly, using the right variable, on relevant pages, at the right time.
A/B testing software – the easy way to test
Using an A/B Testing software is a great way to find out which site changes will determine customers to make the desired actions on site. In an ever-changing environment where websites have to keep up with consumers’ preferences and ways of using the internet, tests should be conducted regularly.
When using the right A/B testing tool, experiments can be simple to create and to implement, without any fastidious changes in the content of the website or in the website’s source.
A/B Testing tools enable you to obtain high-value information about customer behavior and to take the right decision when it comes to changing an element of the site.
Beginner’s guide to A/B testing: 7 unavoidable steps
First of all, a few preliminary considerations:
- The hypothesis should imply a goal achievement. For example, you can create a hypothesis that states “Addressing directly to customers through the website’s content will increase conversion rate up to 5%” or “Removing one field of the contact form will increase leads with 4%”.
- Experiments must be conducted under strict rules. For the results to be accurate and precise, there are a few factors to be taken into account:
- Both page versions must be displayed to the website’s visitors at the same time, for the same period of time. Otherwise, results could be influenced by seasonality, trends and other external events.
- Traffic must be split equally between the different versions of the page. You can choose different segments of users to apply the test to and make changes that would be relevant only to these users.
- The variation pages should run for a period of time that enables the test to give accurate results. Relevance is determined by the total number of visits, the number of visits included in the test and the settled goal.
1. Gathering data. Your analytics tools will provide you with insights into the relevant areas of improvement. Check out the areas on your website that bring the most traffic- they will help you collect data faster. Then look for pages with low conversion rates and high drop off rates. The cart abandonment rate or bounce rates are a few elements worth considering when you are getting started with A/B testing.
2. Setting a conversion goal. The whole purpose of conducting an A/B test is to improve elements that will drive more customers to accomplish the desired action. As a result, the conversion rate will improve. You can use A/B tests to find a way to maximize different types of conversions:
- Completed acquisitions
- Pageview on a strategic page
- Add-to-cart clicks
- Clicks on different strategic elements
- Requests for quotation
- Digital documents downloads
- Newsletter subscriptions
- Clicks on ads or banners
- Any other strategic action on site
3. Hypothesis. Testing starts with analyzing the conversion funnel and identifying what page needs to be improved in order to lead more customers further in the process of converting. Moreover, you have to identify exactly the element which, once it’s been changed, will make a real difference in customer engagement. A good understanding of customer behavior combined with knowledge in website ergonomics can point to this key element.
4. Create variations. After completing these first steps, the elements that require improvements become clearer. Whether you need to work on the copy, to add more prominent call to action buttons, to simplify the design or to make adjustments to the color scheme, get all these laid out and create the second version of the page you want to upgrade.
5. Test duration: run the experiment for at least one week. Visitors will be randomly distributed to either the control or variation page. Their interactions with these pages will be counted, measured and compared. This will allow you to see which version performs better. Yet, in order for the results to be relevant, you need to keep the experiment running during a reasonable period of time. The statistical significance of the sample and the results is an essential part of the A/B testing process: if you don’t have it, the results may not be of much help, nor worthy of being implemented. There are no standard sample sizes but an A/B testing calculator may come in handy at this point, as it tells you exactly what size of the user sample is statistically significant in your particular case and whether the results are powerful enough. You can use the calculator both for pre-test analysis and for the final evaluation of the test.
6. Analyzing A/B testing results. Once the test is done, your A/B testing platform will offer you the data from the experiment and show you which version performed better. If the reasons behind the users’ choice remain unclear, use qualitative methods (focus groups, social media, customer surveys, session recordings, etc.) to get a deeper understanding of the visitors’ behavior, to be able to pinpoint the exact reasons why they prefer certain elements over others. Introduce the qualitative information into a database so you can come back to it whenever you find it necessary. Update it every time an opportunity for qualitative research arises.
7. Repeat. At this point, you may have identified a winning variable. Or not. Sometimes the results are not conclusive, and you need to start anew, with an improved hypothesis or with other elements to be tested. Don’t worry, it’s natural. Positive results don’t always come from the very first round of A/B testing. Go back to step one and make sure you follow all the steps of the testing method. This time you should have a better understanding of what can go wrong and what should go right, so it’s more likely to identify the best performing version.
Yet your work doesn’t stop once you found the winning variation. Conversion rate optimization is a continuous and fluid process that has to be constantly carried on. A/B testing should be performed periodically for you to reach your KPIs time after time. While a successful A/B test will positively impact your conversions for a certain period of time after it has been performed, these results can fade away in time. eCommerce is a very competitive market and if you want to stay ahead from your competitors, you need to constantly adjust your strategy and improve the customer experience on your website.
Which are the best elements to A/B test: A/B testing examples
- Marketing: prices, offers, call to action
- Copywriting: headlines, copy style, length of text
- Design: layout, colors, background, pictures
Challenges of A/B tests
Due to the fact that A/B testing may involve the creation of very similar pages in terms of content, some SEO challenges may arise.
Nevertheless, if you follow Google’s advice on the matter and you conduct the tests properly, all risks can be eliminated. Also, if you use the right A/B testing tool, necessary precautions will be taken automatically.
If you don’t want your page versions to be perceived as duplicates or very similar content, use rel=” canonical” to indicate that there is only one original version of the page, and that’s the one that should be taken into account by robots.
Also, to redirect traffic to a selected version of the page, use a 302 redirect, rather than a 301. This tells robots, that the variation page is only temporary and it’s probably part of a test.
Why do you really need to do A/B testing?
If the benefits outlined in the first part of this article didn’t win you over, here’s what you should know:
- A/B testing methodologies bring you fast results. The whole process is easy to set up and run. What’s more: it drastically reduces your traffic acquisition costs.
- It is also a very versatile procedure: it can be applied to many areas of your website but also to your marketing campaigns. If you use qualitative research to complement your user testing efforts, the results will reveal more than you actually wanted to learn in the first place.
- While the process in itself is pretty simple, the data you gather is vast and complex. You can use advanced analytics for each variation you decide to test, thus collecting precious insights that can be used in other areas of your business as well.
- As opposed to multivariate testing, A/B tests require less traffic for optimal results. This way, you can get actionable insights quicker.
- You can do it on your own, or you can turn to companies offering A/B testing services if you feel like you lack the time or expertise to carry out the experiments.
Omniconvert – A/B testing made easy
At Omniconvert we believe that no marketing conclusions can be universally applied. Something that worked on one site, may not work on another. Thus, research and testing should be the basis of any marketing strategy. Applied correctly, A/B Testing can help you make design, marketing and copywriting decisions that will lead to more conversions from the same amount of traffic.
Omniconvert is an online website optimizer that comprises, among other important features, a very precise and easy to use A/B Testing tool. By using it, you can change elements of design, content, and calls to action, without being faced with the technical complexity of having to implement changes by yourself.
Due to Omniconvert’s ability to segment traffic based on location, referral, weather, and other behavioral parameters, you can measure the performance of a new page version, optimized for a specific group of users.
A/B testing is a very important step in conversion optimization and a must for every site that wants to engage users in a more meaningful way.
Do you want to enhance your UX, improve your website’s performance, test product features and find new ways to communicate effectively with your potential customers?
Then you can book a demo of our powerful A/B testing engine. you will find more than 40 parameters to mix and match in order get to know who your visitors are, where they come from, what actions they previously performed on your website and many other useful insights that will help you focus on what matters most when on their path down to conversion.