It is a common misconception that you can test your way to conversion optimization. An abundance of testing software, some of which is free, makes the temptation only greater. But there are at least three reasons why testing alone — without first conducting research — is not a good idea.
First, how did you decide which elements to test? For example, you may try several variations of calls to action on your web page, by changing their positions or colors. But how did you choose the positions, or the colors?
First, how did you decide which elements to test?
Second, the mechanics of testing means you need to test for a certain period of time, depending on your traffic and expected improvement. Otherwise you will end up wasting time and, more importantly, revenue.
Why revenue? Because each unsuccessful variation or failed A/B test will cost you half the money you’d otherwise make without testing at all. Testing is based on a proportion of visitors being routed to variations. Thus each failed test means visitors to that test underperformed, which results in a loss of revenue.
And the third reason to conduct research is to establish expectations for improvement. Properly conducted research can enable you to estimate how many more conversions you can expect from a change. By having realistic expectations, you can determine how long the test will last and how many visitors you need to test to have meaningful and statistically significant results.
What Is Conversion Optimization Research?
Conversion optimization research is the set of activities necessary to formulate hypotheses for tests that are grounded in data. Without this, you will base your conclusions largely on assumptions or chance.
For an effective testing program, you need to first know a lot about your visitors: who they are, how they perceive your website, where they came from, what they do there, and what they want and why. You need answers to these questions before you embark on a testing program.
Using the answers to those questions, you can pose valid testing queries, such as “Will visitors convert better if a call to action button had contrasting color?” or “If we enable visitors to buy using their social network accounts, will they convert more?.” Questions like these are called hypotheses and they enable you to formulate experiments with confidence that the variations actually solve the problems you detect.
4 Types of Conversion Research
There are typically four steps to conversation optimization research: technical, heuristic, quantitative, and qualitative.
Technical research. Technical research is the necessary first step in conversion optimization, to ensure that the website is functional and enables all the visitors to view the content. Although it is the most basic, it certainly is not perfunctory and should be done thoroughly. After all, if visitors cannot access the site or read it, there’s no reason to optimize their conversions.
Important issues that hinder conversions and that can be uncovered by technical research are cross-browser and cross-device compatibility, broken links, different resolutions, and loading speed. By solving those issues, you increase the visitor base and create more chances that visitors will actually convert.
Heuristic research. Heuristic research ensures the website conforms to best practices and principles of user interface design. For example, studies — most notably by ConversionXL — confirm that “prototypical websites” convert better. Practically, this means that if you sell clothes, for example, your site should resemble other similar clothing sites.
Heuristic research also ensures that the site has a logical structure, to enable visitors to smoothly navigate the site and fulfill their goal — i.e., find the product that solves their problem and purchase it.
This step will frequently uncover problems that may need experimenting to find the best solution. For example, you could uncover that a form is too long and visitors will not fill it out. You have multiple options to solve this, from shortening the form to making it more clear how much time user needs to complete it.
User testing — submitting the website to random users to see if they can navigate it successfully — is a common tool for this part of research. UserTesting.com offers this service, for example.
Quantitative research. Quantitative research relies upon metrics, such as number of visitors, conversion rates, number of visitors who fill out forms, and the number that view videos. This part of research uses tools such as Google Analytics, Adobe Marketing Cloud, or Kissmetrics to track visitors and report their activities on your site. By analyzing and comparing different patterns of visitor behavior, you can uncover what parts of your website are popular, what content engages visitors, and what leads visitors to become customers.
There are many supplementary tools that track only certain aspects of visitor behavior, such as tracking mouse movements, scrolling, and filling out forms.
Quantitative research will supply real data, to understand the activities of your visitors. Quantitative research can support other research and hypothesis. It can also uncover conversion problems, to devise solutions.
Take, for example, a conversion funnel, the path a visitor follows to become a customer. By tracking a properly structured conversion funnel and spotting where visitors drop out, you can improve that step in the funnel, resulting in more sales.
Qualitative research. Finally, there is qualitative research. If quantitative research tells us what visitors do, qualitative tells us why they do it. The best way to understand the decisions of visitors is to ask them. Feedback from customers and visitors can be obtained directly by surveys, interviews, and questionnaires — and indirectly by using reviews, chat logs, or support call logs.
Qualitative research can uncover trust and credibility problems. It can improve content, to conform to the audience. Most importantly, qualitative research makes it possible to create personas, which are hypothetical profiles of ideal customers.
Surveys are by far the most common method of qualitative research. This could be, for example, exit surveys, triggered when a visitor abandons cart. Less frequently used are traffic surveys, which are aimed at all visitors as they navigate the website and are triggered by certain actions, such as clicking for more information or details.
Once you complete the research phase and gather as much data as possible, it is time for hypothesizing.
Hypothesis is necessary to create a viable test proposal. It should determine the target testing audience, a problem the test solves, the scope of solution, and expected or projected result. Those variables allow you to prioritize hypotheses, create tests, and calculate how long they will take.
Prioritizing hypotheses and formulating tests are the final phases of conversion optimization. Both can follow only from properly conducted research.