Burst
- Product Designer & Researcher
- 33 weeks · 2021 to 2022
- Solo thesis project
Burst is a thesis prototype for newcomers to Canada who are rebuilding their social network. It uses a location-based feed so people can ask nearby questions and answer in their own language.
9
40
3
A hundred-page guide can tell you how to settle in. It can’t give you someone to ask.
The federal guide for newcomers runs past a hundred pages and, when I began this thesis, was available in English and French. If you couldn’t comfortably read either, the starting point was already difficult to use.
Language was the obvious problem: 58% of survey respondents named it as a barrier. But 75% also named making connections.
“I find that it’s not easy to befriend Canadians that were born here.”
Nobody I spoke to was short of information. They were short of people they trusted enough to ask. Many relied on groups from their own country, which helped with immediate questions but did not always expand their network.
This one’s personal. I come from an immigrant family, and I grew up hearing how much of getting established here came down to who my parents happened to meet.
Who I was designing for
Newcomers aged 20 to 40 who arrived for work, study, family or safety. Some came alone; others arrived with families. What they shared was the need to rebuild a network while learning how everyday life worked.
Four surfaces, built for someone who doesn’t know anyone yet.
Onboarding
Coach marks and tooltips, broken into short steps. Somebody I interviewed asked for exactly this: “some kind of itinerary, like day by day, first day do these things.”
Anyone arriving is already carrying more than they can hold. Onboarding either adds to that or takes some off.
Post feed
Most social apps assume you already know someone. Burst starts with location instead.
Burst shows what people nearby are asking, so the first interaction can be answering a question before finding a friend.
Every post carries a translate control. You write in your language, somebody answers in theirs.
Notifications use green for attention states, keeping them distinct from errors and warnings.
Community support
Enter a postal code and the app lists the community organisations near you. Document help, or somebody who speaks your language, both happen in person.
The online part is for finding the door. The rest happens in a room.
Messaging
Text, images, doodles, voice notes, GIFs. Playful on purpose.
When your vocabulary runs out, drawing still works. Somebody with fifty words of English can still send a picture of the form they’re stuck on.
Blocking and reporting are in here too, and they were in the wireframes from the start. I’ll come back to why.
I began with refugees. The ethics review changed the project.
Refugees were the group I initially wanted to design with. The faculty ethics review would not approve direct recruitment of a protected population for a solo student project.
I rescoped to newcomers who were not classified as vulnerable. COVID closed the community-centre route too, so I recruited online and worked from digital versions of the arrival guides.
The interviews created one more constraint. When someone lost a thought in English, I simplified the question instead of asking them to push through it. The language barrier was shaping the research while I was trying to understand it.
How I ran the research
Nine interviews, forty survey responses, and a literature review
Three subject matter experts who work with newcomers every day. Six newcomers, picked to spread across the reasons people come: work, study, family, escape. Then a survey of 40 people to test whether the interviews held up.
What the process looked like:
Transcribed with AI, then printed and coded by hand
Affinity diagram to separate what mattered from what was incidental
Journey map across discovery, preparing, landing and establishing
Stakeholder map to see everyone a newcomer actually touches
What the literature said
A study of 12,000 immigrants measured four things. Friendship wasn’t one.
The Longitudinal Survey of Immigrants to Canada followed the same people at six months, two years and four years after landing. It tracked employment, education, healthcare and housing, and showed all four improving over time.
It never asked whether anybody made a friend. That gap pushed me to look beyond information access.
What already existed
Everything I found was in English or French
The options I reviewed were limited, and each relied on English or French.
The sources I reviewed rarely offered information in another language. That friction started the day someone landed.
What the interviews kept returning to
Three themes, and they all led to the same place
I transcribed every interview, printed them out, and coded them by hand. Three themes survived that process.
Community
Making friends came up repeatedly. One person told me your network is your net worth. Those connections were how people practised a language and heard about work.
Finding work
Work came up every time, because work is what lets you stay. Language gates it, so weak English means low-paying jobs taken to survive, which leaves no time for anything else.
Where to go
Newcomers get caught by things natives never think about. The yellow cord on a bus tells the driver to stop. When do children start school, and how do you enrol them?
Work, language, friendships and questions all came back to community.
Four stages, and the one nobody designs for.
Discovery, preparing, landing, establishing. The first three are well covered. Embassies, forms, airport services, a SIM card, a SIN number. There’s a process for each one and somebody whose job it is to run it.
Establishing is where it thins out. That’s work, housing, and building some kind of life outside your own front door. It’s the longest stage by far, and the one with the least behind it.
Journey map from the original thesis · Open to view full size
I spent months planning a pamphlet. The interviews killed it.
For most of the research phase, my answer was a booklet. Translated, broken into small pieces, with top-five lists for everything. Five language apps. Five things to know about taxes. Five about healthcare. I had the whole structure written.
Then I read back what I’d actually found. Most practical questions were answered somewhere online already. What people still lacked was someone they trusted enough to ask.
I’d spent months designing a better version of the thing that already wasn’t working. So the pamphlet went. Burst shifted toward questions, nearby people and community organizations where someone could get an answer from a person.
The three concepts
Booklet, website, app
I took three directions far enough to judge them properly, then killed two.
Why the other two lost:
A booklet duplicates what the government already publishes, and it’s static the moment it’s printed
A website assumes a computer, and plenty of people arrive with a phone and nothing else
An app puts it on the device every participant already used
What made the prototype
What the prototype covered, and what I left out
I built a scope table so the prototype stayed focused.
In the prototype:
Post feed, messaging and translation
Drawing, photos, voice notes and calling
Profile creation and local services
Deferred: a helper rating system, and forums split by topic
I tested the three flows that could break trust.
Eighteen wireframes first, then hand-drawn screens slotted into a 3D-printed phone shell so I could watch somebody hold it instead of imagining it.
Three sessions, each on a flow where failure ends the product. Sign-up, because nobody gives you a second chance at a first screen. Messaging, because connection is the whole premise. Reporting and blocking, because an app that puts strangers together and gives you no way to remove one is dangerous.
What came back was that the tab bar appeared and disappeared unpredictably, which made people hesitate. And that the app wasn’t showing its own core. I’d spent the effort on details and buried the thing that made it different.
The edge case I found while sketching
Somebody could use this to target newcomers
Halfway through wireframing I understood what I was building. A location-based feed full of people who just arrived and don’t know what’s normal here yet is an efficient tool for anybody who wants to take advantage of that.
Blocking and reporting moved out of the settings screen and into the main flows. That’s why they were in the wireframes from the start.
An app for meeting strangers is worthless with no strangers on it.
Burst has a cold start problem it can’t design its way out of. A feed with nobody posting doesn’t answer anything. It only works past a density threshold, in a specific neighbourhood, which makes launching it a distribution problem before it’s a product one.
Two other things would decide it. Whether people answer honestly, because there’s no way to check advice from a stranger. And whether Canadians turn up at all, since the premise depends on people helping somebody who can’t repay them.
What I’d have measured
The numbers that would tell me whether the premise held
Each one tests something specific. If it came back wrong, I’d know which decision was wrong.
What I’d watch, and what would prove me wrong:
Answer rate on posts. Questions sitting unanswered would mean the premise fails
Time to first reply
How many conversations cross cultural groups, which is the entire point
Translation usage, and whether people write in their own language at all
Report rate, and whether reports get acted on
Whether anybody moves from the feed into messaging
Three things I’d pick up first.
- A helper rating system, so somebody could walk a newcomer through an office visit and build a reputation for doing it well.
- A partnership with the government, so the app arrives alongside the paperwork instead of competing with it.
- Proper WCAG compliance, which I scoped and didn’t finish.
I went looking for a language problem. What I found was a connection problem.
The app is not the result. The research finding is: people had information, but not always someone they trusted enough to ask.
The literature tracked employment, education, healthcare and housing. My interviews kept returning to connection, something those measures did not capture. If I had stopped at the literature review, I would have built the pamphlet.
I assumed the answer was information
Months spent designing a better handout
Translated, shorter, easier to scan. It took reading my own interview transcripts back to notice that nobody was struggling to find information. They were struggling to find somebody to ask.
I planned around people I couldn’t reach
The scope was built on refugees before I checked whether I could interview them
Rescoping cost me weeks and narrowed who I could design for. It also made the project deliverable inside one academic year, which I didn’t appreciate at the time.
My first prototype hid its own idea
Polish on the edges, core buried
I’d polished onboarding and left the feed looking like any other social app. Somebody opening it cold couldn’t tell what it was for. The fix took less work than the polish had, which is the part that stung.
Thank you.