Conducting good and thorough research is one of the most critical things within conversion optimization — good and thorough research results in better hypotheses and, therefore, more successful experiments.
But how do you conduct good research? And which sources should you consult? That is covered in this article.
Hierarchy of evidence
Let’s start with the hierarchy of evidence.
The hierarchy of evidence is used in science to rank the strength of results obtained from research. Higher in the pyramid means less risk and, therefore, a higher quality of evidence.
Translated to conversion optimization, the pyramid looks like this.
Unfortunately, we still see that many experiments are based on gut feeling (this should actually be at the very bottom of the pyramid). In principle, there is nothing wrong with that. For example, it can work very well to test colleagues’ ideas to enthuse them for experimentation. But if you are a bit more advanced in this field, you want to set up your experiments based on proper data and user research.
Research with the 6V model
At Online Dialogue, we use our 6V model to conduct our research. The image shows what the 6Vs stand for:
Let me explain them briefly, one by one.
Value answers the question: “What is important from the company’s point of view?”
Before you start doing research, you must know the organization’s mission, strategy, and goals. By aligning with the business objectives, you achieve more impact on an organizational level and with your A/B testing.
We, therefore, investigate the other 6Vs based on the company objectives.
Versus answers the question: “What are our competitors doing, and what is happening in the market?”
Thorough competitor research gives you an idea of where you stand as a company, but it can also inspire things you can test on your website. Make sure that you test all the findings and do not just take them for granted. You may have a different target audience, or your competitor might not validate everything they place on their website.
View includes the entire data research part. What insights can you gain from web analytics and web behavioral data?
You can think of various funnel analyses, analyzes per device, or landing page analyses.
In addition, data visualization is also part of View — for example, a heatmap analysis.
In Validated, we answer the question: “What insights are validated in previous experiments?”
We try to look for patterns in the test results and perform meta-analyses. These meta-analyses often provide many valuable and new insights, especially when this was not well documented before.
Verified, for me, is the most underrated and most interesting source of research. With Verified we look at what is already known about the market, target group, or product in scientific research.
With every assumption we have and every hypothesis we write, there’s a chance this has already been studied in science. And these articles provide very useful insights for your experiments. Check out Google Scholar for articles related to your market and product. If you put in a few hours, you will undoubtedly gain new insights.
Voice embraces all user research, and there are many sources you can use.
Possible sources are polls, surveys, usability tests, and customer service. But there are many more to use to gain lots of valuable insights.
Combine your insights
When you have researched all 6Vs, you have a long list of insights. In general, the best test ideas come from combining different sources.
For example, in your data analysis (View), you see a high exit rate in the shopping cart. You then apply an exit survey and analyze what happens during usability tests (Voice). From science (Validated) and competitors (Versus), you get extra ideas for possible solutions. Based on that, you come up with new test ideas.
From research to meta-analysis
The 6V model helps to gain valuable insights and determine test directions. Based on these findings, you will write your hypotheses. Next, you validate these hypotheses through A/B testing (called ‘randomized controlled trials’ in the hierarchy of evidence).
It is essential to realize that you cannot confirm or deny a hypothesis with one A/B test. Several experiments are required for this. And that’s your meta-analysis.
In short, research with the 6V model helps with the first important insights and test directions. You then validate the insights from the research with multiple experiments.
But keep in mind, your success starts with proper and thorough research.