Capstone project — GA Melbourne · 2023

Ember — bushfire prep that starts in October, not on the day

Preparedness advice for bushfire season is scattered across CFA pages, PDFs and Facebook groups — and most people in high-risk areas start in late December, which is another way of saying too late. Our question: could preparation become a series of small, scheduled ten-minute tasks?

Role: Research & design — team of 3
Timeline: 4-week capstone
With: Residents + brigade, Kinglake VIC
Platform: Mobile app concept

Phone showing the Ember preparedness checklist on a wooden table with a map — Ember concept — seasonal checklist on mobile
designed for patchy reception and shaky hands

The problem

The information exists. Victoria's fire agencies publish thorough, well-maintained guidance — and yet every season begins with most households unprepared. We assumed the gap was awareness.

It wasn't. In week one, a Kinglake resident told me: "I know exactly what I should do. I just can't make myself start, because starting means thinking about it." The problem wasn't information. It was avoidance — and avoidance is an emotional design problem, not a content problem.

Research & personas

We drove up to Kinglake and Hurstbridge twice: six resident interviews, then a Saturday workshop at the local brigade with paper worksheets and dot-voting. Doing this at a community hall table instead of over Zoom was the single best decision of the project — people told us things with a cuppa in hand that no form would get.

Community workshop around a hall table with worksheets, dot stickers and a hand-drawn map — Brigade hall workshop — dot-voting on what actually gets done
week 3 — should have been week 1

The prepared procrastinator

Rob, 54 — Kinglake, 2 acres, one dog

“Every year I say I'll do it in November. Every year it's Christmas.”

Knows the checklist almost by heart
Gets overwhelmed by the size of 'prepare your property'
Responds to small, dated, specific tasks

User flow

01
Property profile
Three questions, no account needed
02
Your season plan
Checklist spread across Oct–Dec, not one wall of tasks
03
This week's ten-minute task
One job. Timed. Finishable.
04
Fire-danger-day mode
Offline-cached essentials, huge type, no decoration

alert mode: big tap targets, works offline, zero flourish

The whole concept rests on one trade: we ask for ten minutes a week in spring so that summer doesn't ask for everything at once. The flow is built to protect that promise — nothing else gets to compete with "this week's task".

The app should feel calm precisely because the topic isn't.

Wireframes

Sketchbook spread of hand-drawn Ember screens with orange highlighter and crossed-out ideas — Sketchbook — the map-first home screen, before we killed it
so much of this page is crossed out. correct.

Our first wireframes opened on a live fire-danger map — it looked impressive and tested terribly. Residents didn't want ambient dread on their home screen; they wanted to know what to do this week. The map moved behind a tab, and "this week's task" took the front door.

Hi-fi & prototype

Two phones showing the final Ember app in burnt orange and cream — Final screens — checklist and fire-danger-day mode
burnt orange: urgent without screaming

The palette is deliberately warm — burnt orange and cream rather than warning red. Testing showed red-on-everything read as "emergency happening now", which spiked anxiety and, perversely, made people close the app. Calm visuals kept people in the flow.

Open the Figma prototype

Usability testing

Five participants in Kinglake, moderated, on their own phones — including one session standing in an actual driveway with patchy reception, which taught us more than the other four combined.

F01The word 'evacuate' poisoned onboarding
Asked early to 'plan your evacuation', two participants disengaged entirely — one handed the phone back.
FixRewrote to 'plan to leave early' and moved it out of onboarding entirely. Engagement recovered in the next round.
F02Offline wasn't a nice-to-have
The driveway session: pages hung, the map never loaded, and the participant said what everyone was thinking — 'this is exactly when I'd need it.'
FixCut the live map. The season plan and day-mode checklist became fully cached, readable with zero bars.
F03Progress bars backfired
'14% prepared' read as a guilt score, not motivation. One participant: 'great, now I feel behind on fire.'
FixReplaced percentages with streaks of completed weeks — forward-looking instead of accusatory.

Outcome & reflection

This was a capstone, so the honest outcome is a tested prototype and a hard-won lesson — not a shipped product or a metric. The brigade asked to keep the worksheets, which remains my favourite piece of feedback ever.

What I'd do differently: co-design with the brigade from day one. We treated them as expert reviewers in week three when they should have been in the room while we framed the problem. Every finding that mattered — avoidance, language, offline — was visible in that first conversation. We just hadn't had it yet.

What stays with me: designing for an emotional state is a real, learnable craft. Fear makes people close apps. Calm, small, dated steps make them come back.

the user isn't irrational — the fear is rational. design around it.

Next project

01Myki Mate

worth a look →

Rethinking the myki top-up so it survives a moving tram, one hand, and a low balance warning.