Designing applications for spatial computing is a whole different beast, and good user research isn’t just a nice-to-have step. It’s the foundation for making something that’s actually intuitive and engaging, an experience that goes beyond what we’re used to on a 2D screen.
Key Takeaways
- Go into the real world and watch how people move and interact in physical spaces *before* you even think about designing your digital overlays.
- Use hardware with specialized recording capabilities, like a Varjo XR-3 with its built-in eye-tracking, to capture the subtle interaction data you’d otherwise miss.
- Iterate fast and cheap with low-fidelity prototypes, even physical mock-ups, to test spatial layouts and figure out if your user flows make any sense in 3D.
- Get users to think out loud during testing. This qualitative feedback is your best tool for finding out where the cognitive load is too high or where the experience causes spatial discomfort.
- Dig into the spatial telemetry data, gaze vectors, how objects are handled, user movement, to fine-tune your interaction models and make the environment feel responsive.
1. Define Your Research Objectives and Hypotheses
Before you start collecting any data, you have to know what you’re trying to learn and what assumptions you’re testing. For spatial computing, that means getting specific about interaction models, user comfort over long sessions, or whether your UI elements actually work in 3D space. “Understand user behavior” is way too broad and won’t get you anywhere. A much better hypothesis would be something like, “Determine if direct hand manipulation of virtual objects reduces task completion time by 20% compared to controller-based interaction for users aged 25-45 in a simulated assembly task.” Getting that specific forces you to pick the right methodology and gives you something you can actually measure.
Pro Tip: Your research objectives should be framed as questions you can answer with data you can observe. With spatial apps, you need to be asking about things like motion sickness, how aware the user is of their surroundings, and the cognitive load of being in a 3D environment, which are all problems you just don’t run into with standard 2D apps.
2. Choose the Right Research Methods
You can’t just use your old UX research playbook for spatial computing. Screen recordings and surveys alone won’t cut it. You need a mix of traditional methods adapted for 3D and some newer techniques. Contextual inquiry becomes incredibly important, actually going out and observing users in the environment where they’ll use the app. Imagine watching a mechanic use an AR overlay to fix an engine. How do they tilt their head? Where are their eyes looking for information? Usability testing is still a core activity, of course, but your test setup has to account for the physical room and the fact that the user is physically present and moving within the experience. Ethnographic studies can also uncover pain points that people would never think to mention in a simple interview.
Common Mistake: Thinking you can get by with remote testing tools built for websites and 2D apps. These tools almost always fail to capture the physical interaction data that is so fundamental for spatial computing, completely missing the nuances of depth perception, proprioception, and how users react to haptic feedback.
3. Select Appropriate Hardware and Software Tools
The tools you pick will make or break the kind of data you can get. For the qualitative stuff, you’ll need a solid video setup that can record both the user’s physical movements and what they’re seeing inside the headset, which gives you the full picture of their interaction. Some high-end devices like the Varjo XR-3 even come with integrated eye-tracking, a goldmine for data on where people are actually looking, what draws their focus, and how they scan a scene. For AR work, you’ll obviously need a device with AR capabilities, like an Apple Vision Pro or a Meta Quest Pro, to run realistic tests. And don’t forget the software, specialized tools like Unity’s XR Interaction Toolkit or Unreal Engine’s XR tools are designed to help you log interactions, movement paths, and object manipulations.
On the quantitative side, the telemetry data coming directly off the device is pure gold. This means positional tracking data (x, y, z coordinates), rotation data (pitch, yaw, roll), and all the interaction events like button presses or recognized gestures. Make sure your development environment is set up from the start to log these events with precise timestamps. For example, a simple but effective setup in a Unity project is to write custom scripts that log the transform position and rotation of the headset and controllers every 100 milliseconds, writing it to a file alongside any custom events like “grab” or “release.”
4. Recruit Representative Participants
The people you test with have to be the people you’re building for. It seems obvious, but it’s a common trap. If you’re making an architectural visualization tool, go find architects, not just your tech-savvy friends. You have to consider their prior experience (or lack thereof) with spatial computing, their age, technical skill, and even physical things like hand dexterity for gesture controls. A more diverse group of participants will always expose a wider range of problems. You should aim for a minimum of 8-10 participants for qualitative usability testing, which, as Jakob Nielsen’s research has shown for years, is usually enough to spot the most common issues.
Pro Tip: Make sure your recruitment includes a mix of total novices and experienced spatial computing users. The beginners will immediately trip over fundamental design flaws you’ve become blind to, while the power users can give you feedback on more advanced interactions and point out performance issues.
5. Design Your Test Scenarios and Tasks
Your test scenarios have to be realistic and tied directly to the research questions you defined earlier. If you’re building an AR app for remote assistance, a good task would be having the participant walk a remote expert through a simulated repair using your AR overlay. If it’s a VR training module, the tasks might involve working through a complex virtual factory floor or performing a delicate simulated procedure. It’s best to break down big, complex interactions into smaller, more focused tasks. For every single task, you need to define what success looks like ahead of time (e.g., “the user correctly placed all 5 virtual components within two minutes,” or “the user found the virtual meeting room without asking for help”).
As you’re designing these tasks, always be thinking about cognitive load because spatial computing can be mentally exhausting. Don’t throw a dozen new concepts or complex, simultaneous interactions at a participant right at the start. Begin with simple tasks and ramp up the complexity. And please, always run a pilot test of your scenarios with your own team first. It’s the only way to find confusing instructions or technical bugs before you waste a real participant’s time.
6. Conduct the Research Session
When the session starts, your first job is to make the participant comfortable in a space free of distractions. You’ll need to explain the study’s purpose, but be careful not to give away your specific hypotheses, as that can easily bias their behavior and feedback. The most valuable thing you can do is get them to “think aloud” while they’re using the app. This stream-of-consciousness commentary is where you’ll find their real thought processes, their frustrations, and those small moments of delight. Record everything you possibly can: a camera on the user, a screen recording of what they’re seeing in-app, their audio commentary, and all the back-end telemetry data. Pay attention to their body language too, because a slight head tilt or a frustrated sigh often says more than words can.
For AR apps, you have to make the physical test environment match the real-world use case as much as possible. If it’s an app for a warehouse, you better be testing it in a big, open space that feels like a warehouse. For VR, double-check that the user has plenty of room to move around safely and give them very clear instructions about their physical play space boundaries before they put the headset on.
7. Analyze Your Data
This is where all that raw data gets turned into something you can actually use. For your qualitative data, you’ll want to transcribe the think-aloud recordings and start grouping observations into themes like “navigation difficulty,” “UI element confusion,” or “comfort issues.” The goal is to find the patterns that show up again and again across different participants. Your quantitative data, like task completion times, error rates, and gaze data from eye-tracking, can be run through statistical analysis. But don’t just look at the numbers. Visualizations like user pathing in 3D space, heatmaps of gaze data, or scatter plots of where people tried to interact can reveal problems that a spreadsheet would hide. For instance, a gaze heatmap might show that five different users all looked everywhere *except* the corner where you hid a critical button.
One of the most powerful things you can do here is to connect the qualitative and quantitative data. When a user says, “That part felt really hard,” and you can look at the telemetry and see that their head movement spiked and task time doubled for that specific action, you’ve got a much stronger finding. This triangulation of data gives you real confidence in your conclusions.
8. Iterate and Refine
User research isn’t something you do once and check off a list. It’s a cycle. You take your analysis, identify the biggest pain points, and prioritize them based on how bad and how common they are. Then you go back to the drawing board to develop design solutions or tweaks. You build a new prototype with those changes baked in and then you go right back to testing with users. This loop of research, design, and testing is absolutely fundamental to making good spatial computing applications. For example, if your first round of tests shows that everyone hates a particular gesture, you might prototype an alternate gesture or even add a voice command as a backup for the next test round.
This iterative approach is invaluable. In a field like spatial computing that’s changing so fast, what seems right today can be obsolete tomorrow. Getting that constant feedback from real people is the only way to keep your app relevant and genuinely usable.
Solid user research for spatial computing requires a multi-modal approach that combines careful observation, hard data analysis, and a commitment to refining your work over and over. By really getting a handle on the unique demands of 3D interaction and human perception, you can build spatial experiences that are not just clever tech demos, but are actually useful and enjoyable for people. This iterative mindset is exactly what’s needed to solve the new mobile UX challenges appearing in this field.
What’s the biggest mistake people make in spatial computing user research?
The most common mistake is just taking old 2D research methods and trying to apply them directly to spatial computing. They don’t account for the unique problems of 3D, like physical movement, a user’s sense of space, or how comfort changes over longer sessions.
Is eye-tracking data really that important for this kind of research?
Yes, it’s incredibly important. Eye-tracking gives you objective proof of where users are actually looking, what grabs their attention, and how they scan a 3D scene. This is a goldmine for figuring out where to place UI and how to structure information.
Should I bother with physical prototypes for a spatial computing app?
Absolutely. In the very early stages, physical mock-ups (even with cardboard and foam) can be a cheap and fast way to test things like spatial layouts, object scale, and how users might move through a space. It can save a ton of development time down the line.
What specific telemetry data is worth collecting from a spatial app?
You’ll want the position and rotation data for the user’s head and their controllers. You also need to log all the interaction events, grabs, button presses, gestures, as well as task completion times and any custom events that are specific to your app’s goals.
How many users do I need to test with for spatial computing?
For qualitative usability testing where the goal is to find the most common problems, 8-10 participants is a good rule of thumb. If you’re doing a quantitative study and need statistically significant results, you’ll need a much larger group.
