Stori Referrals.
How a single insight from six user interviews — "users feel ashamed to refer" — reframed a leaderboard into a mission system, and grew New Active Bankers by 22.85%.
A digital bank scaling fast in an underbanked market.
Stori is a venture-backed fintech on a mission to democratize credit access for the 400 million underbanked consumers in Latin America. Operating in Mexico with global teams in Arlington (Virginia), Mexico City, and Asia, Stori has become one of the country's leading digital banks — surpassing 2 million applicants for its credit card product since launch.
I joined the Growth Squad as the embedded product designer. Growth squads are accountable for one thing: moving acquisition and activation metrics. Every screen we shipped had a target — and we were measured against it.
Referrals weren't growing the bank.
The existing referral program was underperforming. Participation was low, drop-off was high, and the campaign wasn't moving the metric we cared about most: New Active Bankers (NABs) — users who actually open and use the credit card.
Two patterns were obvious in the data:
- Only a small percentage of users referred more than 10 people. Power users carried the program.
- Many users abandoned the experience the moment they realized they wouldn't reach the top of the leaderboard.
The brief from the Product Lead was clear: redesign the campaign to grow NABs. But the data told us what users did — not why they hesitated. That's where I started.
A small squad, a high-stakes brief.
I worked closely with the Product Lead and a Business Analyst on this initiative. The Business Analyst owned the quantitative picture — funnel data, cohort behavior, payout economics — while I led the qualitative side and the design end-to-end.
My responsibility was full-cycle: frame the real problem, validate it with users, design the new experience, and ship it in a way Engineering could build and Marketing could activate. I was the only designer on the project.
Two stages: what they do, why they hesitate.
I designed a two-stage research plan to complement the existing analytics. First, quantify the pattern; then, understand the emotion behind it.
Stage 01 — Quantitative (Maze survey)
A short survey distributed in-app to size the problem at scale. The goal was to validate hypotheses around awareness, friction, and intent — not to discover something new yet.
Stage 02 — Qualitative (6 user interviews)
Semi-structured 30-minute interviews with active and lapsed referrers. The single, focused goal: understand how users feel when they're asked to refer.
Working hypotheses going in:
- Users feel ashamed about referring.
- Users don't have enough people to refer.
Key insights
The interviews surfaced seven insights. I categorized each as a positive signal, a negative signal, or an opportunity area to make the synthesis legible to the wider squad.
"Users refer when they feel they're helping — not when they feel like salespeople. The leaderboard turned a generous act into a competition."
The old experience optimized for the wrong behavior.
The previous program was framed as a competitive leaderboard. Higher rank → bigger prize. On paper, it worked: clear incentive, gamified, easy to communicate. In reality, it broke down for two reasons that the research made obvious:
Competitive leaderboard
Users earned rewards based on their rank against everyone else. Top spots won iPhones, PlayStations, large cash prizes. Lower ranks won less.
Why it broke: participation collapsed the moment users realized they couldn't reach the top. The framing was zero-sum. And the "earn money for referring" mental model amplified the shame surfaced in research.
Mission-based system
Users complete missions tied to the act of helping — not to outranking others. Each mission unlocks an instant, individual reward. There are no losers.
Why it works: every user has a path to a payout. Effort feels personal, not competitive. The story shifts from "win against others" to "help your friends, get rewarded."
This was the inflection point of the project. The research didn't just validate hypotheses — it changed the brief. The problem wasn't UI friction. It was the underlying mental model of the program.
From competition to cooperation.
The new flow was built around three principles, each tied directly to a research insight.
Principles
- Helping framing. Copy and visual language reposition the act of referring as helping a friend access credit — not selling them a product.
- Individual progress. Every user sees their own mission ladder. There is no leaderboard, no comparison to others.
- Instant, predictable rewards. Each completed mission triggers a payout. No waiting until the end of the month, no rank-based uncertainty.
The flow
Seeing the two experiences side by side makes the reframe concrete — the same goal, the opposite mental model.
Overview. The referral program was designed as a competitive experience where users could earn rewards based on the number of successful referrals. A leaderboard showcased user rankings, with higher positions granting better prizes.
Problems. Only a small percentage of users referred more than 10 people — and many users, upon realizing they wouldn't reach the top ranks, became demotivated and disengaged from the program.
The metrics moved.
Within the first month after launch, the redesigned referral program delivered measurable lift across all the metrics the squad cared about.
Beyond the numbers, the qualitative signal mattered too: post-campaign surveys showed a measurable increase in user satisfaction with the referral program. Users reported feeling more comfortable inviting friends — exactly the emotional barrier the research had surfaced.
What I'd take into the next project.
What worked
Investing two weeks in qualitative research before opening Figma. Six interviews don't sound like much, but they were the difference between optimizing the wrong thing and reframing the brief. The mission-based system would never have surfaced from data alone.
What I'd do differently
I'd push earlier for an A/B test against the old experience instead of a full cutover. We have strong directional evidence the new flow drove the lift, but cleaner attribution would let us isolate the impact of the framing shift versus the visual redesign.
What I'd explore next
Personalized incentives based on user segments (high-network referrers vs. tight-circle referrers behave differently), and longitudinal tracking to see whether mission-driven referrers retain better than leaderboard winners — a hypothesis I'd want to validate before we ship the next iteration.