Wallet Activation Drop-Off Reframing

A team was tasked with reducing drop-off in Grab’s e-wallet activation.

Data showed ~20% drop-off at phone number verification. The assumption: How do we improve verification?

That assumption went unchallenged. I questioned the premise itself.

Users’ phone numbers were already verified:

• During Grab account creation

• Through repeated usage across services

The verification step existed because it had always existed.

Action

• Proposed removing the step entirely

Impact

• Eliminated the 20% drop-off

• Reduced SMS costs

• Removed unnecessary engineering complexity

What this demonstrates

• Systems thinking over surface optimization

• Willingness to challenge “industry standard” logic

• Bias toward irreversible simplification

Timeline Negotiation Under Pressure

A team faced an immovable regulatory deadline.

Even after:

• Adding headcount

• Cutting scope to MVP

…the timeline was still impossible. The default response was overtime.

Instead, I engaged the compliance team, the group that actually understood regulatory priorities.

Insight

Several “must-have” MVP features were irrelevant to regulatory approval.

Action

• Re-scoped the submission to focus only on regulator-critical requirements

• Deferred non-essential flows (e.g. discovery journey)

Impact

• Met regulatory timeline

• Preserved product quality

• Avoided burnout and rework

What this demonstrates

• Understanding power dynamics

• Leveraging institutional knowledge

• Solving for approval paths, not internal narratives

Cash-Out Fee Controller (Problem–Solution Fit)

Business requested a feature to incentivize cash-out fees, effectively encouraging users to remove money from the platform.

On investigation, the real goal was different: Encourage Grab drivers to spend within the Grab ecosystem (OVO)

The proposed solution was inherited logic from a previous cash-in fee model.

Action

• Challenged the solution, not the goal

• Redirected budget toward targeted spending incentives instead

Impact

• Aligned incentives with actual business intent

• Avoided building a counterproductive system

What this demonstrates

• Strategic interrogation of requests

• Ability to say “this is the wrong tool” without derailing momentum

Illustration vs Error Icon Debate

Another team pushed for a generic error icon to “improve clarity.”

My team pushed back, not for consistency’s sake, but for emotional calibration.

Principle

• Visuals communicate emotion before information

• Overly harsh signals increase anxiety and abandonment

Resolution

• Framed the decision around user psychology, not design preference

• Used an illustration that preserved clarity without amplifying negativity

What this demonstrates

• Judgment under ambiguity

• Emotional intelligence applied to systems design

• Knowing when consistency matters and when it doesn’t

Research Automation with AI

Research at Grab had become:

• Too slow (4–6 months)

• Too expensive (USD 40k–60k per study)

• Too inconsistent to support product decisions

Designers were doing research work they weren’t trained or resourced for.

There was no engineering support. No official roadmap.

Action

• Evaluated multiple AI engines against vendor reports

• Built a validation model (accuracy, consistency, relevance)

• Defined clear guardrails for human vs AI research

• Deployed using internal AI platforms as MVP

Impact

• Research cycles reduced from months → minutes

• Designers freed from research mechanics

• Adopted organically, then formally supported

• Became standard practice across design teams

What this demonstrates

• High-leverage system design

• Shipping under extreme constraints

• Turning unofficial work into organizational standards

This isn’t a collection of stories. It’s a repeatable pattern.

When logic alone doesn’t move systems:

• I interrogate the premise.

• I identify the real decision-maker or constraint.

• I redesign the path to make forward motion the default.

The outcome is consistent: Progress without waiting for permission.