How To Interpret Quantitative UX Metrics
Article written by Huyen Hoang
Huyen Hoang is a User Experience Researcher at Codelitt. Codelitt helps companies create better product experiences for their users by designing and building people-driven solutions with the speed, technology, and innovation of a startup.
For many researchers working in UX, quantitative data analysis can cause a lot of anxiety. UX researchers provide interpretations that can impact businesses with said data, which can be an overwhelming amount of pressure. There are many reasons UX researchers are often overwhelmed by quantitative metrics. One of the biggest reasons is that many people find statistics intimidating. Another reason is that most UX professional programs, such as boot camps and online certificates, don’t go in-depth on quantitative research methodologies. Generally, data science programs more thoroughly explore these topics. Therefore, this article will provide a general overview on how to interpret quantitative UX Metrics.
First, let’s identify some common pitfalls to avoid when interpreting quantitative UX Metrics.
Correlation is not causation. Two UX metrics may be correlated, but that does not mean one caused the other. A third phenomenon may be responsible for one or both of the two UX metrics in question.
Confirmation bias. Quantitative UX metrics can be misinterpreted when a researcher is looking for patterns in the data to support a hypothesis they want to be validated, while simultaneously rejecting data patterns that would invalidate that hypothesis.
Irrelevant data. There is so much data that can be collected for a given product. It’s essential to be clear on what kind of data should be prioritized to avoid analyzing and interpreting data that is not relevant to UX.
Small sample sizes. It is not good to gather a lot of irrelevant data. It is also not good to collect relevant data from too few users. Generally speaking, quantitative data analysis aims to make predictions about a population with a margin of error. Having a small sample size will produce less reliable results.
This list is in no way comprehensive but covers the most common mistakes in analyzing quantitative UX data. Avoiding these pitfalls ensures your interpretation of quantitative UX metrics is objective.
Plan for high-quality and relevant quantitative UX data.
Proper interpretation of quantitative UX metrics starts before gathering any data. There are overarching questions that practitioners need to ask to keep on track and make sound interpretations.
Some questions to consider are: What are the goals and objectives of the quantitative research you are gathering? What research questions are attempting to be answered with quantitative UX metrics? What methods will be used to interpret data? Who are the stakeholders who will use the data?
Investing the time to define and answer these questions allow UX researchers to focus on highly relevant metrics to goals and objectives.
Focus on UX-related metrics and not business metrics.
There can be an overwhelming amount of metrics for business analytics. So the first step is to narrow it down so that time isn’t wasted focusing on irrelevant data to UX.
Pro tips: understand UX Metrics versus KPIs.
UX Metrics are quantitative data used to measure, compare, and track users’ experience interacting with a digital product over time. These are associated with user behaviors and attitudes. KPIs (key performance indicators) are quantitative data used to measure, compare, and track the overall goals. These goals typically are tied to revenue, growth, retention, and user counts.
It is essential to focus on UX data that aligns with your goals and objectives for research.
Have a streamlined data wrangling process in place.
A critical part of the quantitative data interpretative process is ensuring data is reliable before analyzing and leveraging it for insights. At this junction is where data wrangling (the process of discovering, structuring, cleaning, enriching, validating, and publishing the data) comes in. This process can be very lengthy and time-consuming.
Data professionals spend as much as 80% of their time preparing data for analysis. UX professionals cannot afford this much of their time to be sucked up in cleaning and organizing data. But suppose your research operations have streamlined processes for how to wrangle data. In that case, this saves a lot of time and removes the risk of gleaning insights and making interpretations from incomplete, unreliable, or inconsistent data.
Use storytelling to communicate findings.
Data visualization is an art. And explaining data visuals is a craft. Not many can do these two things well. This is why storytelling is such a powerful skill. Graphs and charts are great, but if a researcher cannot tell a story to explain the data, the findings have minimal impact on business decisions. Additionally, people, including business leaders, are moved by stories.
It is essential to know how to choose the right data visualization type. Generally, there are four goals for data visualization types: 1. showing relationships, 2. showing distribution, 3. showing the composition, or 4. making comparisons.
Asking the following questions will help you define the best visualization type for the right audience:
- What is the story you want to tell?
- Who is the audience you want to tell the story to?
- Do we want to analyze trends?
- Do we want to demonstrate composition?
- Do we want to compare two or more sets of values?
- Do we want to show changes over time?
- How will we show UX Metrics?
Once these questions are answered, it becomes easier to decide if a pie chart, a line chart, a spider chart, a bar chart, or a scatter plot is the best visualization type to tell the user experience story.
Synthesize your insights and draw valuable conclusions
Now comes the moment where the synthesis of quantitative UX metrics data serves as a change agent for the user experience. Extract facts from the data. Remain objective by being aware of the pitfalls previously discussed. And make interpretations of the data. The goal is to generate valuable recommendations.
Good recommendations are:
- Constructive. They offer a solution rather than focusing on the problem revealed by the data.
- Specific. They identify wherein the user experience recommendations are most applicable.
- Actionable. Suggestions should be active. Use language that is active rather than passive to inspire change.
- Concise. Plenty of recommendations can be generated from any given set of UX data, but not all of them will significantly impact the user experience. Prioritize the most important ones.
- Measurable. Good recommendations can be measured so that there can be evidence a change has occurred and an impact has been made.
- Balanced. Identify both the strengths and weaknesses.
For UX practitioners, the volume of quantitative data available in today’s digital world is vast. And correctly interpreting quantitative UX metrics can be a daunting task. While it’s worth investing in highly technical skills, often, it’s more about processes that enable sound interpretations of UX metrics. The key is to remain objective, focus on relevant data, have simplified procedures for data cleaning and analysis, tell a good story with said data, and draw valuable conclusions to improve the user experience. Interpreting quantitative UX metrics is more about the process than sophistication in statistical knowledge (some tools take care of this). The goal is to have simplified, focused, and repeatable processes.