CASE STUDY 01 · PAYPAL CONSUMER · 2019–2024

Early-Life Journeys
at Scale

How I redesigned the way PayPal's global consumer product moves users from sign-up to meaningful financial habit — by treating the problem as a systems failure, not a UI one.

6

SURFACES COORDINATED

200+

MARKETS SERVED

~40

EXPERIMENTS LED

THE CHALLENGE

Users recognized PayPal. They didn't know what it could do for them.

Quantitative research showed strong feature recognition — users could name PayPal's products, pass survey benchmarks, demonstrate brand familiarity. Qualitative research told a different story. In real-task sessions, the same users scrolled past visible features without registering them.

ON THE CORE DESIGN CHALLENGE: We could teach users to do one more thing, but we couldn't help them understand the system they were now inside. The experience produced activation without accumulation.

The conventional fix — more tooltips, better copy, a smarter onboarding flow — addresses the symptom. I went after the structure. What was needed wasn't a better funnel. It was a system that sustained learning over time, across surfaces, without forcing it.


THE SYSTEM I DESIGNED

Not a funnel. A reinforcement model.

The key insight was that learning after a successful action is more durable than learning before one. I shifted investment toward completion states, post-transaction detail, and contextual reinforcement — the moments where understanding actually crystallizes.

Early-life surface map — where understanding crystallizes or decays
Key moment
Decay risk
Signal passes forward
Onboarding
Intent signal
Identity verify
First value moment
Signal passes forward →
Home
Personalized layout
Activity peek
Module discovery
Growth moments
Wallet
Add instrument
Verify & link
Error recovery
Ledger
Transaction history
Detail drill-in
Contextual nudge
⚠ Decay without reinforcement
Reboarding
6-month check-in
Contextual re-introduce
Never same rec twice
Core insight
Learning after a successful action is more durable than learning before one. Completion states — not entry flows — are the highest-leverage teaching moments.
What connected it
Intent signals from onboarding informed Home layout, Wallet sequencing, and Growth recommendations. Six siloed surfaces sharing state.

Intent signals traveled

What we learned during signup—likely use case, instrument type, geography—informed Home layout, Wallet sequencing, and Growth recommendations. Surfaces that had never talked to each other started sharing state.

Completion states as classrooms

The moment after a first successful P2P transfer is the strongest educational moment in the product. The user just succeeded. They're receptive. That's when we introduced what was adjacent, not before.

Reboarding, not just onboarding

A ~6-month check-in framed as a tune-up, not a sales pitch. Rules: never the same recommendation twice. Never two consecutive recommendations. Designed to feel like a knowledgeable friend, not a notification.


MY ROLE

The connective tissue across six surfaces

Home, Onboarding, Wallet, Financial Snapshot, the Activity Ledger, and Growth were each owned by different product teams: different PMs, different roadmaps, and different success metrics. My job was to make them behave like one system.

I established the cross-surface governance, the shared state logic, the experiment libraries, and the review cadences that let those teams coordinate without constant design bottlenecks. I partnered with Data Science to build the behavioral models that determined what appeared, when, and for whom. I worked with Legal and Compliance on every experiment that touched credit, disclosures, or regional rules.

I also stabilized the design organization through two CDO transitions, building the documentation and decision frameworks that kept teams moving when leadership context evaporated.


OUTCOMES

What it produced

2×+

Improvement in push notification opt-in through consent-first proxy design

Multi-feature activation in early cohorts through intent-aware personalization

Durable

Framework that outlasted two leadership transitions and remained the org's reference point

WHAT MADE IT LAST

The system was designed to be modular. When dependent features didn't ship on schedule, we removed those surfaces without dismantling the model. What survived was the logic — and it remained extensible for teams who came after.