← All worknow showing ( 09 ) / 24

The moment before the mistake

The moment before the mistake key visual

a way for dbs customers to report fraud and lock their accounts themselves, as it happens. soon after we finished, dbs announced a kill switch much like ours.

Client DBSDuration 1 month, 2022Role researcher, designerTools figma, figjam
( 01 ) Challenge

scams were rising, and reporting one was slow.

singapore recorded 23,931 scam cases in 2021, 5,020 of them phishing. banks were racing to go digital and fraud kept pace. we asked two things: how does a bank know a customer is real, and how does a customer report something while it's still happening?

23,931scam cases in 2021
5,020phishing scams
$633mlost to scams
( 02 ) Research

we started with how fraudsters work.

we read forums and reddit threads from people who'd been scammed. younger people are warier, but scams play on instinct, so age protects less than you'd expect. i did competitor analysis later than usual this time. banks mostly leave disputes to agents, while neobanks like revolut are raising expectations.

( 03 ) What users wanted

in most cases, instant help wasn't instant.

banks taking more responsibility for recovering lost money( 01 )
a person to talk to in a crisis, because a bot has no urgency( 02 )
control over what happens on their own account( 03 )
( 04 ) Personas

we cut one.

we started with a digital native and a digital novice, but couldn't learn enough about the native's banking to find real pain points, so we dropped it. the journey map followed david before, during and after a dispute. the chance to win him back starts the moment he realises something is wrong.

( 05 ) Design studio

we sketched together, remotely.

the research was the constraint. we took turns sharing screens and critiquing each other. it left us with practical questions: where do new features go so people notice them, and which icon actually reads as chat?

design studio sketches
design studio sketches
( 06 ) What we built

we added four features to the app dbs already had.

new device alerts, so you can approve or deny a login from elsewhere( 01 )
a wallet for card actions: activate, report lost, change pin, manage overseas spend( 02 )
a kill switch that suspends every account and card at once, one tap from "more"( 03 )
a digibot with clear prompts and a way through to a live agent( 04 )
kill switch: find it, review, confirm
( 07 ) Testing

we critiqued twice a week, on real phones.

most feedback was about content. keep banking copy simple, and write buttons the way staff would talk to you at a branch. also, the 'bot' in digibot was a dealbreaker.

( 08 ) Afterwards

dbs shipped a kill switch.

dbs later announced anti-scam measures due by 31 october 2022, including a kill switch that freezes all banking activity. we'd designed ours as more than a button, down to how it feels to use when you're panicking. we first wanted to redesign large parts of the app. narrowing to fraud gave us a sturdier foundation.

( ✎ ) looking back

if i did it again

( 01 )a/b test information layouts across different levels of tech literacy
( 02 )map the attacker's journey, so the bank puts controls where they do the most good
( 03 )map the existing service, so we could design routes in for people who usually get left out
( 04 )study fraud trends and think like a criminal when designing backend fraud analytics
( 05 )get legal, risk, business and product agreeing on one vision before designing anything
road not taken

i spent a while on chatbot design: intro messages, mostly, and how a bot should handle someone who's upset. there was far more to it than we had time for, and testers had already told us the 'bot' in digibot put them off.

( ↗︎ ) Afterwards
Next track →
Designing for people who already gave up