Whether you want to redesign your digital product or create a new one from scratch, the UX research and discovery phase is fundamental when learning more about users' pains and gains. Research hypothesis helps you validate your assumptions and guides you through the whole process. It can discover what should be the next actionable steps in the design process, based on scientifically-proven statements and facts.
Table of contents:
What is a hypothesis in UX research?
A hypothesis in UX design is a statement about how you think users will behave and what kind of solutions will fit and respond to their needs. In other words, a hypothesis defines how you think something works based on your research, knowledge, and experience. You can use this tool to guide the rest of your user research process and find a factual answer you can support with numbers, and appropriate probes with users whose actions, reflections, and feelings are tested and analyzed.
A hypothesis is a testable form of an assumption that, through experiments, can be proved or disproved. One can test it with experiments using qualitative or quantitative research methods. Defining a hypothesis helps you determine what data you need to collect from your users and how to interpret it.
A hypothesis formulates how two or more variables are related.
- "Elderly people use mobile devices less often than young people do."
- Or: "If we make our website accessible for people with low vision, they will have a better experience on our site."
Hypothesis for redesign and new product design
If you implement a new design to an old one, created hypotheses can relate to how you expect the user's behavior will change. In this case, you can formulate a hypothesis by asking questions like:
"What do I expect users to do once they have access to this feature?"
For instance, if you think about adding to your online learning platform app the feature of downloading courses for offline mode, you can ask questions like:
- How do learners usually use offered courses?
- Where and when do they prefer to learn?
- Are there additional values for students to download offered classes and materials?
- What would students do when they had an opportunity to download courses?
Revised assumptions and tested even in the primary way can help you formulate a research hypothesis that can look, for instance:
We believe that creating the feature of downloading courses by XYZ app students will increase their engagement in taking classes and finishing them earlier.
If you offer the user a completely new product, a hypothesis will relate to more fundamental issues, and you can formulate it through questions like:
- What are the most effective ways of online learning?
- What motivates people to do new courses and finish them?
- How do people search for new courses and classes?
- What encourages people to learn new skills?
Those questions can help you formulate a research hypothesis that usually should be based on previous research on this topic and then lead to the next steps. The hypothesis formulated in this case would sound, for instance:
We believe that certificates given to students of courses on the XYZ platform will encourage them to take classes and complete them.
How to structure the hypothesis?
To formulate a hypothesis, you need to clearly understand the problem you are trying to solve and what your users want. The best way to find a research hypothesis is by brainstorming with a team.
1. Formulate research questions
When we see a problem, it is natural to immediately try to find the solution. This works the same in the product design. Therefore, before you jump directly into solutions, it is crucial to consider what you don't know and for which issues you don't have answers to. Note assumptions as well. Write them down, individually or together with your team during the workshop. Categorize them later, organize them into groups and prioritize.
Then, focus on the most important and ask yourself:
"What do I think is the best solution for this problem?".
To formulate research questions and turn them into assumptions, you can use the How Might We technique by IDEO. This technique helps to frame the question in an open-ended way without imposing answers. For example, using the previous model, you want to encourage students to finish their courses on your online learning platform successfully. So, you can ask the question:
How might we encourage students to finish the course once taken successfully?
2. Structure research hypothesis
Once you have prepared a prioritized list of questions, you can formulate a research hypothesis. It combines a prioritized set of questions with well-thought-out solutions. The solutions that we will list now can be based both on experience, previous research done by you, or by other people. Try to think about how you would answer the asked questions.
If the question is, “How might we encourage students to finish the course once taken successfully?”
In our example, we can answer, for instance:
- Giving certificate after completing the course
- Providing free access to the software and books related to the course
- Offering 1:1 mentoring with the professional after completing the course
With the question and solution combined, you can formulate a hypothesis. This way, you can eliminate assumptions and clearly define the question and answer important from your product and project perspective.
You can formulate it in several ways, but the most common and profound is a format of hypothesis statement proposed in the book Lean UX by Jeff Gothelf and Josh Seiden:
We believe [this statement is true]. We will know we’re [right/wrong] when we see the following feedback from the market: [qualitative feedback] and/or [quantitative feedback] and/or [key performance\nindicator change].
Jeff Gothelf and Josh Seiden
This form shows that the hypothesis has to be testable and the data that constructs it, measurable. But, you can also create a hypothesis that is linked to the features. In this case, you need to collect information as:
- The business outcomes you are trying to achieve - The users you are trying to service - The user outcomes that motivate them - The features you believe might work in this situation
Jeff Gothelf and Josh Seiden
It is a measure of your business success. You have to define what you want to achieve. Is it:
- more visitors to the website?
- more sign ups for the newsletter?
- encouraging people to have discussions with each other?
Create a list of possible outcomes, and check which one has the most prominent result you seek.
You have to remember that you are not an end user. You should not think about yourself as a user to avoid wrong assumptions. Users chosen for the hypothesis should be based on the proto-persona or persona, defined carefully during the brainstorming with the team.
Designers should be a kind of empathetic advocates of the user. In creating hypotheses, they should think about what would be the outcome for the user. They can find out that by asking questions such as:
- What are our users trying to accomplish?
- How do they want to feel during and after this process?
- How can we help them reach their goals?
This is one of the most popular parts of the designer's work, as they often can work directly on solutions and features. It is important to remember that designers should also focus as well on:
- research and investigation of the problem
- who our users are,
- what would be the tangible outcome for us as a company
- what would be the outcome for the users.
We often start with solutions; however, the order should be different. Together with the team, we should analyze what feature would be crucial and help to achieve user outcomes.
By putting together previously collected data and insights, we can start to structure the research hypothesis accordingly to this template:
We believe this [business outcome] will be achieved if [these users] successfully achieve [this user outcome] with [this feature].
Jeff Gothelf and Josh Seiden
In the analyzed context of the online learning platform, we can formulate, for example, this hypothesis statement: We believe this growth in the number of platform class participants will be achieved if students complete the course with the certificate approved by well-known professionals and companies.
3. Testing hypothesis
In UX research, you should test every hypothesis and, this way, validate it or invalidate it. We always should check if the hypothesis statement formulated by us is right or wrong. If you want to test whether your hypothesis is correct, you need to conduct experiments to help you answer this question: "Does what I think is happening actually happen?"
Conduct an experiment to test your hypothesis. Ensure you follow the scientific method by collecting empirical and measurable evidence to obtain new knowledge. Outcomes should be measurable to determine whether your hypothesis has succeeded or failed.
There are many different ways to test your hypothesis. You can both use:
- Quantitative research methods like surveys, usability testing, A/B testing
- Qualitative research methods like in-depth interviews focus groups, ethnographic research, diary accounts
Combining quantitative and qualitative methods can help us get more reliable results and ensure we revise the hypothesis in the best way possible.
Qualitative insight helps us to understand the emotional aspects of product design. It provides the “why” to give context to the quantitative “what” insights provided by analytics tools. It gives us a sense of what’s driving the behavior and provides guidance for design improvements that improve the experience. This makes our customers as well as our business more successful. By balancing qualitative and quantitative insights, we are using data to inform rather than dictate our design decisions.
Jeff Gothelf and Josh Seiden
We can also formulate when we assume that the experiment is valid. To do it, we can use other kinds of formats, for instance:
We will run X studies to show more information about students (experience, education, previously finished courses, motivations, needs, frustrations), and ask follow-up questions to identify the students emotions associated with the learning process (difficulties, grit, pleasure, engagement, goals etc.). We will know the hypothesis is valid when we get more than 70% identifying the certificate as an important motivational factor for learning.
The next step is to run appropriate tests with chosen research methods, prepare their scenarios and scopes, set a timeframe, formulate questions, and recruit users.
Once you conduct the test, you should be able to check if your hypothesis was true or false. With this knowledge, you can start to implement checked and validated hypothetical solutions or reject false ones. In both cases, there is always the possibility to formulate new hypotheses that can be tested because the design is a never-ending process.
Hypothesis-driven design process
The hypothesis-driven design process is a user research method that helps product teams focus on the most critical areas of their product. It involves testing a hypothesis about how users interact with a product and then iterating based on the results of that test. The goal is to keep refining the hypothesis until what needs to be done next is clear to ensure your product meets its users' needs.
The hypothesis-driven design process starts with identifying a problem you want to solve. This could be something like:
- "How can we make this website easier for our customers?"
- or "How can we improve engagement on our app?"
Next, you'll need to gather data from users about how they currently do things so that you can compare them with how they would do something if you implemented your solution. After collecting this data, you can test hypotheses about which changes will significantly impact user behavior by creating prototypes and then collecting feedback from users who have tried them out.
After each round of testing, follow up with more interviews and other research methods until you find out what people need from your product or service. This way, you'll know exactly where to focus your efforts in future iterations of development!
What are the benefits of research hypotheses in digital product design?
There are many benefits of research hypotheses, among others:
- They allow you to make better decisions about the direction of your product because you're basing those decisions on accurate data about people's actual needs and behaviors.
- You can validate or invalidate your assumptions.
- They help you evaluate your designs before implementation.
- They allow you to test features as they're being developed.
- While designing a new product, you can check if the target audience would be interested in it and its proposed features.
- You can reduce risk and increase the certainty you have in assumptions and understanding relative importance.
The value of the research hypothesis is that it provides a framework for user research to follow. It provides direction for what you're looking for and shows you how to interpret your findings.
Usually, we already propose the solution when we see the problem. This works as well in the case of digital products and their users. Based on our experience, we often believe that we know everything about our users. But we are not the end user of the designed product, and often we can be surprised that our solutions don't work and don't respond to other people's needs.
Hypotheses in UX research and hypothesis-driven design process show that it is essential to slow down, hide our assumptions, beliefs, expectations, experience, and ready answers for everything, and conduct deep research into what we, in reality, don't know.
After testing hypotheses, you can build a clear path and see scientifically-proven steps that allow us to achieve our business goals while providing the best user experience that responds to the real needs of our customers.
Gothelf Jeff and Seiden Josh, Lean UX, 2016
Kahneman Daniel, Thinking, Fast and Slow, 2013
Lai Sylvia, 5 steps to a hypothesis-driven design process, 22 Mar 2018
Levitt Debbie, UX Research Without a Hypothesis and for Products That Don’t Exist Yet, April 2015
Hypothesis statement, Uxspot.io
Lenneville Christie, How to write a strong hypothesis, GitLab
Holliday Ben, Everything is hypothesis driven design, Sep 27, 2017
Was it helpful?