03/04/2017
AB testing as a way to increase conversion rates in eCommerce
There is a wide debate being led concerning website optimisation and multivariate testing as a way to improve the conversion rates on any given website. Many people are still following “the best practices” that have been devised a few years ago and are still the mainstream options in these operations. There are some who propose new ways and advocate that the widely accepted best practices are not sufficient. The confronted opinions include the one claiming that there is a flow that needs to be followed by the T, and the other that claims that the entire A B testing process should lean heavily on personalization and varies from case to case, thus rendering “best practices” useless.
The only way to deal with a messy situation like this one is to actually dissect the entire problem and try to find out the right answer. That is exactly what we are going to do in this article!
In order to understand the entire problem better, and to obtain the starting points for discussion, it is important to define the right segments that should be considered.
• Global or local AB testing
• Demographic influence on the results
• A B testing in relation to personalization
• AB testing in relation to eCommerce software
• AB split testing in direct marketing
Global or local approach?
In most cases, the opinion you will hear is that the AB testing should be carried out in a global way. This is one of the most impactful reasons why many really nice looking and effective eCommerce stores end up being a total fluke!
The global trends are important, and there is a possibility to lead global campaigns if the markets your eCommerce store covers are similar. Still, without deep localization when it comes to the AB testing, the results will be far from satisfactory. Simply, potential customers definitely come from different backgrounds, and they appreciate being addressed in the way they will understand easily. This stands not only for the language option, as the most obvious one, but to all other segments as well. People from Germany are not thinking the same as people from the UK or US!
What needs to be done: You need to carry out conversion rate optimisation in a way that will show certain progress, and to do it in a way that can provide measurable results!
How: The best way to do it is to compare to what you already have and what works well. One of the options in AB testing should be based on the website variant you already have for global approach. After that, optimise it for the specific market and include all the local options (payment, shipping, customer support). The split test with 50-50 approach will give you the exact results concerning that specific geography.
Useful tip: Scout the opposition and consult the experts. If your digital agency has some experience in the local market you wish to pursue, they will be able to propose the best approach to it. Also, see what your direct competitors are doing, and scout the most successful ones. That will save you a lot of time and money!
Demographic variables
This is one of the most complicated steps in multivariate testing. People do not react in the same way to the same version of the website, and it is in most cases directly related to their sex and age. There is a lot of reason to this occurrence. Younger generations are usually more wired onto the global trends and they are easier to reach. Also, men, women, and adolescents do not react in the same way to the same visual outlays. In order to make sure that your AB testing results are relevant, you always need to compare them to the demographics involved. There is a way to do this on a more advanced level, using personalization tools, but that segment will be discussed in more details further down the road. If you are doing it without any personalization tools, which is the most common case, the only thing you need to take care of is comparing the results to the actual visitors. Just as an example – if your target market consists of women, you should completely exclude visits made by men, and then compare the results.
What to do: Optimize the very results of multivariate testing!
How: Only compare the relevant results. Google Analytics can help with this, as it provides detailed insight in demographics of the visitors that were involved in split testing. Always compare them by groups and you will get reliable data.
Useful tip: Always use a sufficient number of test samples and try to cover them not only by numbers, but also by time of the visit. Usually, well converting customers have a specific time window when they are most active. Use the experience from the global campaign to determine peak periods, and try running tests during that period. Always keep the time difference in mind!
A B testing in relation to personalization
Personalization tools could be the best option there is to reach out to every single customer in the best possible way. These tools track the behaviour of visitors, and are capable of giving them the suggestions that will lead to higher conversion rates. The efficiency of these tools can be best defined through multivariate testing. There are options that could allow you to run one version of the website with the personalization turned on, and the other without it.
These services are usually quite pricey, so identifying their real value is something that could benefit your efforts on a huge scale!
What to do: Make a larger set of data to compare using all the resources you have at your disposal!
How to do it: Try running a full set of personalized AB tests. Use all website variants you have prepared and run them both with and without the personalization tool running. That will give you a double set of data compared to the standard approach. Use all possible comparisons to see what generates the best conversion rates, and then scale up the results on a monthly level. Then compare the benefits to the investment personalization tool requires and you will have a pretty good idea where to head further.
Useful tip: If there is an option of using a personalization tool that is easy and cheap to integrate, and allows sign up for a very short period of time (either via demo option or via short term contract) – that is the option you wish to pursue. This approach gives great results with minimal investment and minimal risk of committing long term to something that is not producing results.
Multivariate testing in relation to eCommerce platform
Some of the best eCommerce platforms out there have AB testing as a native feature. Still, it is not the case with all of them. Some of them show problems in multichannel trade which involves deep localization for different markets. Some others have trouble with running on multiple markets using multiple currencies. There are a lot of possibilities to hit the bump on the road if you have chosen eCommerce platform that can’t produce all you want to do. Also, the reporting options are not the same with all of them. Some provide deep data that can be easily used for analysis and manipulation, while the others don’t have sufficient depth in reporting. Make sure your platform can support the effort you want to make!
What to do: Make sure your eCommerce platform is compatible with your AB testing campaign. Make sure you will obtain the data that could be used for drawing conclusions!
How to do it: Adjust the testing options to your platform. Use only the options that will give reliable results and then compare the data.
Useful tip: Consult your digital agency about these issues. They know exactly what can be done! They have done this in the past and have all the experience they need! Discuss all options and only then make a decision on what to do and how to proceed!
AB testing in direct marketing
Many people make a huge mistake by applying split testing only to their storefronts. Although this approach is well justified, and that is the most important step of the entire conversion rate optimisation process, there are other segments that simply have to be covered. One of the most commonly overlooked things is direct marketing.
Namely, a lot of businesses run direct marketing campaigns in a standard and global way. This means they have a standardized NL (newsletter) that is then only translated in different languages. Although translation is important, it is far from everything you are able to do in order to increase conversions through this channel. A bit of an effort in this segment can create some serious benefits!
Since this part of the AB testing is completely independent from the eCommerce platform you are using, it requires minimal investments and minimal efforts, and can produce incredible results!
What to do: Try to split test the newsletters using different approaches and determine what works the best!
How to do it: First and most importantly – scout the opposition. It is easy to subscribe to their newsletters and track what they are doing. It will give you plenty of ideas on how to proceed. Also, consult with your web agency. Devise an entire set of newsletters that include various approaches, and run periodical campaigns using different layouts. It will give you the exact idea which design works the best, and which approach provides best results.
Useful tip: Exclude seasonal campaign newsletters from this test. If there is a seasonal sale going on, the conversion rates will be higher than usual so you will not obtain accurate data. Try to run the split tests under the same circumstances. Additionally, you might target different demographic groups with different sets of newsletters.
If you follow all the mentioned steps in a proper way, you should be able to bring your online business to an entire new level. Well tested solutions WILL convert well, providing increased conversion rates, increased sales, and increased profit. Of course, it is the process that takes time and resources, but the final result is something that is definitely worth all the hassle.
All being said, it is clear that the “best practices” in split testing are present, but far from a complete solution. There are so many variables that it is next to impossible to put everything into any kind of guide. Even all the variables we have mentioned in this article are just general guidelines, and there is a lot more to the entire process if you want to be successful and invest your time and money in a proper way.
The only right approach is to make a personalized plan for your operations. This can be done only through communication with your digital agency.
We at Studioworx have been through numerous similar projects. What our experience so far taught us can be summarized into this – only the personal approach to every client, resources at his disposal and his needs can produce the results that are in line with our high standards. In order to devise the proper strategy, we need to collect all the relevant data, from targets and goals, via capabilities, to the budget. It can be done only through the constant communication with the client and constant involvement of both parties into the process.
In this way we are able to produce results ALWAYS, with no excuses!
If you would like to know how we could help your business run all the necessary AB tests and achieve your business goals, simply give us a call or contact us via our website! We will be there for you to analyse the project with you, propose solutions, and help you carry out every single step of it!
We do not compromise with the quality! Call us and see for yourself!