Rethinking a "social robot" brief through the lens of real user friction, and proving that good UX beats novelty tech.
This project started with a brief I found genuinely exciting: explore how emerging technologies, and specifically "social robots," could improve everyday university life. We chose to focus on something deceptively mundane: the university canteen.
Trento and Rovereto's university canteens are functional, affordable, and central to student life. But they're also built entirely around implicit local knowledge: menus in Italian only, unwritten queueing etiquette, a payment system with unadvertised rules, and staff who often don't speak English. For an Italian student who grew up with this system, none of it registers as a problem. For a newcomer arriving without that shared context, every single step becomes a small negotiation with uncertainty.
That gap is what made international students the ideal lens for this research. They have no implicit knowledge to fall back on, which means every point of friction in the system surfaces immediately and honestly through their experience. If a service works for someone with zero prior context, it will work for everyone.
Our objective was to move past assumptions about what technology could do, and instead build an evidence base, grounded in real behaviour, real language, and real frustration, that could tell us what technology should do, if anything at all.
Within a team of six, I took ownership of two connected strands of the research: facilitating our focus group and leading the qualitative cross-coding and synthesis that tied our three research methods together.
I designed and ran a 90-minute focus group with seven participants at the San Bartolomeo student residence, restructuring our guideline from direct questions into broader conversational prompts. Deliberately sequencing the "social robot" prompt last so participants surfaced real frustrations organically first.
I cross-coded transcripts I hadn't conducted myself, forcing a more objective, pattern-driven reading. I then led the process of pulling recurring codes from observations, interviews, and the focus group into a single, coherent set of findings.
Being close enough to the data to facilitate a live conversation, but disciplined enough to code it objectively afterwards: knowing when to listen and when to step back and pattern-match.
We deliberately built a three-tier research funnel rather than relying on a single method, because each method on its own has a blind spot that the others cover.
The result of this triangulation wasn't just "more data": it was confidence. When a friction point showed up in what we watched, what people told us privately, and what a group of strangers agreed on out loud, we could trust it as a genuine systemic issue rather than a one-off anecdote.
When we raised the idea of a social robot as a solution, both in interviews and in the focus group, we consistently met polite but limited enthusiasm. One thread came up again and again: nobody wants another app, another interface, another thing to learn on top of a system that already didn't explain its own basic rules.
We shifted our research question from "how could a social robot improve the canteen experience?" to the more honest version: "how can the canteen experience be improved for newcomers, and what role, if any, can social robots realistically play?"
In short: users didn't reject technology. They rejected novelty technology layered onto an unfixed foundation. The social robot wasn't a bad idea; it was simply premature. That distinction, between "interesting to build" and "actually needed," is the single most transferable insight from this project.
Translating these findings into product direction meant resisting the instinct to jump straight to "cool" solutions, and instead building a prioritised, evidence-backed roadmap that a real product team could act on.
A redesigned digital menu inside the existing Opera4U app: visually grouped combinations, allergen icons, and a running price total to address payment-surprise friction before the till.
Bilingual signage and a consistently structured, English-inclusive digital menu treated as a non-negotiable design system requirement, on the same level as accessibility.
The payment friction, menu confusion, and queue anxiety compound. Fixing the app's reliability, the menu's clarity, and the till's transparency together has a materially larger impact than fixing any one in isolation.
Get the foundational digital experience right first, then treat conversational or robotic interfaces as a second-stage enhancement for specific, well-defined moments rather than a blanket replacement.
This is exactly the kind of evaluation the role demands: taking a brief that assumes a high-tech answer, and having the research discipline to say "the users are telling us something different, and here's the staged plan that respects both what they need now and what might genuinely add value later."
Observing behaviour, asking about experience, and stress-testing ideas in a group each surface a different kind of truth. Learning to only trust a finding once it showed up across all three made me far more careful about the difference between "an interesting anecdote" and "a pattern worth designing for."
Restructuring the focus group guideline mid-process, moving from direct questions to open prompts, and deliberately sequencing the "social robot" topic last taught me that how you ask shapes what you learn just as much as what you ask.
The most valuable output of this was the evidence-backed case for not building one yet, and a clearer, more useful roadmap in its place.
Six researchers each holding a piece of the puzzle meant the real work was in the connecting: reconciling codes, checking consistency, and building a shared narrative the whole team could stand behind.
Ultimately, this project reinforced something I believe should sit at the heart of good product design: technology should earn its place by solving a real, evidenced problem, not by being the most exciting option on the table.