Chinese immigrants in Canada carry clinically significant mental health burdens — and face an app landscape built for someone else entirely. This is a project about designing the moment before someone asks for help. Before they have the words for it. Before they're ready to call it anything at all.
Population figures are drawn from public Canadian sources and document the demographic and access gap this project responds to. The qualitative findings, personas, and walkthrough results presented below are my own primary research — a concept project, not a clinical study.
Every interview I ran started the same way. I'd ask how life had been since the move. And almost every person — regardless of how long they'd been in Canada, how much had or hadn't worked out — would give me some version of the same three words.
一切都好.
Everything is fine. It is not a lie, exactly. It is a social architecture — a structure built from 面子 (face), from 孝顺 (filial piety), from the specific weight of being the one who cannot fail because the whole family made this sacrifice. To say "I'm struggling" in many Chinese immigrant contexts is not just uncomfortable. It is a kind of loss. And so people don't say it. They say 一切都好, or they say nothing at all.
The mental health toll is real and documented: isolation, identity disruption, credential loss, family separation, the specific loneliness of being a financial lifeline for everyone back home. But existing apps weren't built for this. They were built for a user who has already decided they need help — who is ready to name it therapy, ready to pay for it, ready to see their face on a profile. That user and the one I was designing for had almost nothing in common.
This is not a therapy app. It is a project about the entry point — the moment before someone decides whether to walk through a door at all. That moment is harder to design than anything behind it, and almost nobody is designing for it.
I ran semi-structured interviews with 6 Chinese immigrants in Vancouver and Toronto — 2 months to 9 years in Canada. I ran a competitive audit of 5 mental health apps. I spent time in Chinese immigrant WeChat groups and Xiaohongshu communities. I was listening not just for pain points, but for how people talked about difficulty at all — because the vocabulary of distress was the first design constraint.
The table below shows the gap that shaped every design decision: what participants said, and what it actually meant for the product.
I audited five apps, crediting real strengths where they exist: Headspace has a Chinese-market product. Calm has a Chinese language option on iOS. BetterHelp has Chinese-speaking therapists. These are genuine features. The criteria below are specific to what my research identified — not chosen to make 路伴 look good, but to define the actual gap.
Each criterion maps to a specific research finding. Stigma-aware entry → Finding 01 (the naming problem). Anonymous architecture → Finding 03 (face and privacy). Immigrant acculturation content (not just general mental health, but content for the specific stress of immigration) → Findings 02 and 05. Non-clinical mood framing → the check-in iteration. Peer community by default → Finding 04 (WeChat as social infrastructure).
| App | Stigma-aware entry point | Anonymous by architecture | Immigrant acculturation content | Non-clinical mood framing | Peer community (default) |
|---|---|---|---|---|---|
| Headspace Has a Chinese-market app (冥想) | ✕ | ✕ | ✕ | ✕ | ✕ |
| BetterHelp Chinese-speaking therapists available | ✕ | ✕ | ~ | ✕ | ✕ |
| Calm Chinese language option on iOS | ✕ | ✕ | ✕ | ~ | ✕ |
| 7 Cups Closest existing fit overall | ✓ | ~ | ✕ | ~ | ✓ |
| Wysa | ~ | ~ | ✕ | ~ | ✕ |
| 路伴 Lùbàn | ✓ | ✓ | ✓ | ✓ | ✓ |
~ = partial or opt-in only. The gap is not Chinese-language support — several competitors offer that. It is designing all five criteria together, from the ground up, for this specific cultural entry-point problem.
These are composite personas built from interview patterns — not edge cases, but the dominant profiles across my research. Both would describe themselves as fine. Both became the filter every design decision passed through.
Her parents saved for years to fund her education. She tells them everything is going well every Sunday, and mostly she believes it. She hasn't made close friends yet. She spends late nights scrolling Xiaohongshu posts from other Chinese students abroad — feeling both less alone and more so with each one. She has chronic headaches she attributes to "just stress." The possibility that they might be something else has not occurred to her, or if it has, she has not let herself think about it.
He was a respected engineer for 15 years in Shenzhen. In Canada, he works below his previous level while his credentials are being requalified. His wife has started noticing he seems distant. He says he's just tired. He sends $800 CAD back to his aging parents every month — money there is no margin to cut — and he performs capability to his family, his colleagues, and himself with the same discipline he brings to everything. He has no vocabulary for what he actually is. The closest he gets is: tired.
The original brief was: design a better mental health app for Chinese immigrants. The research made clear that this brief was fatally named. No product labeled or framed as mental health would be opened by the people I'd just spent three weeks talking to. I wasn't designing an app. I was designing a door.
The name came before the wireframes — an unusual sequence that turned out to be essential. 路伴 was tested informally with three people from the interview pool. All three said it felt like something they might actually open. 路 means road. 伴 means companion — someone who walks beside you, not a guide, not a doctor. The name doesn't promise healing. It promises company. That distinction shaped every feature decision that followed.
The sequence mattered. Most products name last. I named first — because for this project, a frame with the wrong name attached was worth nothing. Everything else was built on top of whether 路伴 felt like something Linda or David would open.
Every clinical mental health product I audited uses the same visual language: cool blues, clinical whites, rounded sans-serifs, sterile space. I made the opposite choices — not for aesthetics, but because the visual register of "clinical" was itself a trigger I needed to avoid.
These are the structural wireframes from weeks 5–6, before any visual design decisions were made. The goal at this stage was to answer one question per screen: what is the least this screen needs to do? Colour, typography, and warmth came later. These frames were about decision points, flow, and what the app would refuse to ask.
The daily check-in was the hardest screen in the product — it had to feel completely ordinary to someone actively performing normalcy. I ran three distinct versions through informal walkthroughs. Versions 1 and 2 were tested with the first three participants I could bring back from the interview pool; version 3 with the full set of four. That's why the counts below shift from thirds to quarters — same people, sessions added as I could schedule them. Two versions failed. Each failure was more informative than the last.
The distance between "clinical" and "natural" in UI is far smaller than it looks to a designer who didn't grow up in the cultural context being designed for. Small framing choices carry enormous weight. The slider and emoji faces would have been perfectly acceptable in another product. They failed here because of what they implied — not what they showed. Metaphor is not a stylistic choice. It is a structural one.
Every screen decision passes the same test: would someone saying 一切都好 still open this? The product doesn't promise healing. It promises company — and the possibility of more, on the user's terms, at the user's pace.
Home → Journal → Community → Resources. That order is the product's thesis. You land in a daily ritual, not a resource directory. Professional help sits at the far right — present, real, filtered for Chinese-speaking and free/sliding-scale providers. It is never the greeting. It is what the app trusts you to reach for when you're ready.
The whole product is four destinations behind a single low-stakes gate. The entry flow never branches into a "are you okay?" decision tree — it deposits the user in a daily ritual, and lets every deeper action stay optional.
The weather metaphor gets the most attention because it went through three iterations. But the other screens had their own rationale. Here is where each came from.
This is a concept project, so I didn't have production metrics. What I did have were four informal walkthrough sessions with participants drawn from the original interview pool. I gave each participant a prototype and five tasks without guidance. The table below shows what happened, what I heard, and what I changed.
The purpose wasn't to prove the design worked. It was to find out where it didn't — and fix it before the final version was built. Every change in the design traces to a specific observation in this table.
| Task (given without guidance) | Round 1 | Key participant quote | Change made | Round 2 |
|---|---|---|---|---|
| Complete first launch without asking what the app is for | 3 / 4 | "Why does it ask how I feel? That's a bit much right away." — P2, stopped at original first question | Changed first question from "How are you feeling today?" to "How have you been settling into life here?" — reframing as adjustment, not emotional check-in | 4 / 4 |
| Complete the daily mood check-in | 1 / 3 slider version |
"This feels like a form at the doctor's. I'd close this." — P1 on the 1–10 slider. Emoji version: "It looks like a customer service review." — P3 | Slider → emoji (abandoned) → weather icons. Final question: "今天的天气是？" The weather frame required no clinical vocabulary and cleared cultural resistance immediately. | 3 / 4 weather version |
| Find the private journal and write one sentence | 3 / 4 | "Can anyone see this? I need to know before I write anything." — P3, before writing a single word | Made the lock icon persistent and visible in the journal header (not just in settings). Added "🔒 Only visible to you" as a permanent label, not a tooltip. Journal was also moved to tab position 2 (before Community) after 2 of 4 participants navigated there first. | 4 / 4 |
| Read the community feed and say whether you'd post | 4 / 4 found it |
"I'd only use this if nobody knew it was me. Ever." — P4. "I'd feel pressure to respond properly if people could comment." — P2 | Confirmed anonymous-by-architecture. Removed comment feature entirely — replaced with 我也是 (me too) as the only available reaction. Added persistent "Always anonymous" label in the screen header. | 3 / 4 said they'd post |
| Navigate to mental health resources and find a Chinese-speaking counsellor | 2 / 4 | "I didn't want to go to a section called 'Support.' That felt like admitting something." — P1 | Renamed section from "Support" → "When you need more." Applied Chinese-speaking filter by default (not opt-in). Moved free/sliding-scale options to first position in the list. | 4 / 4 2 said "I'd come back" |
Small n, informal walkthroughs — directional, not statistical. One task (willingness to post publicly) moved 4/4 → 3/4 after I made anonymity more explicit; the honest read is that confronting privacy head-on surfaced a hesitation the earlier version had simply hidden.
Not whether the design was good — whether specific decisions held up under real use by the actual target user. Every row in that table is a design decision that was either confirmed or changed. The ones that didn't change (community structure, tab order for journal-before-community) were confirmed by behavior, not just assumption.
The most honest version of a reflection isn't "here is what I learned." It's "here is what I believed at the start that turned out to be false." This project changed three things I thought I understood about design — and surfaced one gap I'd close first if I ran it again.
I entered this project believing that the most valuable thing I could give isolated immigrants was connection to others going through the same thing. So I made community the second tab — right after home. Then two of four walkthrough participants navigated to the journal before the community screen, without prompting. They needed to articulate something privately before they were ready to read other people's articulations. I swapped the tab order. Private expression precedes public witnessing. I hadn't designed for that sequence — I'd assumed the opposite.
I had three alternative names ready: Settle, Space, 寄托. I planned to pick based on gut feeling after the wireframes were done. An interview participant — I was wrapping up, the recorder was off — said "路伴" out loud when describing what she wished existed. Something in how she said it stopped me. I ran an informal test with three people from the interview pool before committing to any wireframe. When all three responded to 路伴 differently than to the alternatives — something physically relaxed in how they talked about it — I understood that naming was doing structural work, not branding work. I nearly made that decision last. It should always have been first.
My initial synthesis said: WeChat is infrastructure, therefore the app must feel like WeChat. I spent a week sketching WeChat-adjacent UI patterns — similar message bubble layouts, similar red badge notifications. It all felt wrong. The real insight from the WeChat finding wasn't about visual familiarity. It was about privacy expectations. WeChat's "Moments" controls — posts visible only to chosen contacts — had given users a precise mental model for selective sharing. 路伴's anonymity-by-architecture works because it matches that expectation. I was designing for the surface when I should have been designing for the trained mental model underneath it.
I'd recruit specifically within sub-communities I underrepresented. My six interviews skewed toward Mandarin-speaking, graduate-student-age women in Vancouver — because that's who was easiest to reach. David's persona is thinner than Linda's because the men I approached were less willing to engage with the subject at all, which is itself a data point, but not a substitute for depth. Masculine emotional expression under acculturation stress is its own design problem. It deserved dedicated research and its own design thread, not a paragraph in a persona card. If I ran this project again, that would be the first thing I fixed.
路伴 would need deep partnerships with community health organisations, Chinese immigrant advocacy groups, and culturally competent practitioners in Vancouver and Toronto. It would need ongoing cultural review — there is no static version of this design that stays correct over time. And it would need to be built, in significant part, by designers and researchers who come from the communities it serves — not just by people who listened carefully enough.