Nudge: a thoughtful pause before Buy Now, Pay Later.

Buy Now, Pay Later turned food delivery, clothes and groceries into one-tap credit for young Malaysians — through SPayLater, GrabPayLater and the like.

The target user was specific: college students, fresh graduates and young workers new to managing their own money. Heavy mobile shoppers who use BNPL because they don’t hold credit cards, and many of whom sit in the B40 income tier — which makes small accumulated debts disproportionately dangerous.

So an impulse that is fast and emotional has to be met by an interface that is just as fast and emotional.

Two things compound: with no cash changing hands there is no felt loss, and most money trackers rely on numbers and charts that young people find hard to keep up with. Both point the same way — speak to System 1, not just System 2.

We built two archetypes from the target group: a student who borrows to keep up, and a young professional who spends to cope.

Their pain points differ, but both trace back to the same gap — the payment no longer feels like a loss. Each need became a mechanism.

Following feedback on the first proposal, we put behavioural theory at the centre: every mechanism answers one specific cognitive bias.

Before drawing a screen, I mapped every flow in FigJam — seven flows, over 120 steps and 16 decision points, from first launch to redeeming a voucher.

Condensed, they come down to three parts: a short way in, one dashboard everything returns to, and a single loop that every purchase has to pass through.

A non-essential purchase is frozen in a block of virtual ice that slowly melts over the cooling-off period. No lecture — just time.

If the timer runs out and you actively choose Abandon purchase, it counts as a successful defence: you earn Willpower Points and keep a No-Spend Streak alive. Resisting has to feel like winning something, not losing something.

The avatar is you, later: debt drains HP and stamina, thins the hair and dims the room, while saving earns materials to upgrade your future self’s shelter.

It replaced an earlier “debt monster pet” idea, and that was my call. Young people tend to see their future self as a stranger whose problems aren’t theirs. A pet is easy to ignore; your own future life getting worse is not. So the avatar is the user’s own future — from cardboard box to cosy studio.

Enter a desired item and the AI becomes a gatekeeping boss that attacks the purchase — you have to defend yourself while it counters with something worse.

The best one-liners become collectible cards — “Premium Dumb of the Year”, “Capitalism’s Favorite Child” — made to be screenshotted and shared. The tone is the design decision: playful enough to wake System 2, never so harsh that it tips into shame.

RM is abstract. Plates of nasi lemak and months of a Perodua Axia car loan are not.

Enter an RM shoe purchase and plates rain down under a physics engine until they bury the shoes. Type the amount, name the thing, watch the reveal, then decide.

I gave each colour a single meaning and held to it across onboarding, the dashboard, the cost visualiser, the future self, the vault and the roast chat, so six separately designed flows still read as one app.

So a screen’s colour tells you how to read it before you read a word: lime when you are winning, blue when the app wants you steady, coral when something is about to cost you.

The project was scoped to be shown, not just submitted: four deliverables went out together.

The hi-fidelity Figma prototype, a full brand identity package (logo, colour, typography, icons, poster assets), a narrative short film, and a campus exhibition booth with banners and a screen mirroring the prototype. Social posts carried the same message to the platforms where the impulse usually starts.

Over the three-day graduate exhibition, about 50 visitors and students used the prototype hands-on and told us what worked and what didn’t — the first time Nudge met people outside the team. That was guerrilla testing: good for spotting confusion and gauging the tone, not a controlled study.

The critique about missed details is the one I take forward: the next pass is a detail sweep across all six flows, so every number, state and spacing holds from screen to screen.

Alongside it comes a moderated test with 8–10 Malaysian BNPL users aged 18–30, focused on one question: is the vault understood as protection, and does the roast tone land as humour? The answer will shape the next iteration.