CASE STUDY · BANCO GALICIA · 2026
A conversion problem that was a measurement problem
Redesign of the bank's package upgrade flow: from an embedded webview to a native experience that first decides who gets to see what.
THE PROBLEM
Official conversion was 6%, but 70% of the people entering the funnel had no available offer. The number measured nothing.
THE DECISION
Offer evaluation as an entry prerequisite and a native 3-step flow.
THE OUTCOME
Flow designed, validated and built. It didn't reach production before I left the bank.
THE DIAGNOSIS
The 6% hid an 80%
The upgrade flow lived in a SPA embedded as a webview inside the app. The experience was inconsistent with the native design system and official conversion was 6%.
That number explained nothing. I processed the customer databases with AI to reconstruct the real funnel and found that 70% of the users entering it had no available offer: they were browsing without being able to buy. Among those who did have an offer, conversion was 20%. Real dropout among eligible users was 80%.
That data changed the project's question. It wasn't how to push 6% higher. It was how to measure real intent and design only for those who could convert.
THE DECISIONS
Four decisions, with their cost
I led the design team end to end: problem mapping, scope, flows and high-fidelity prototypes. I worked with Melanie on content design and with Nico, from the design system team, on custom components validated to scale across the bank. I used AI to analyze benchmarks and explore screen variants before defining the final flow. Technical constraints and business rules came out of the work with product and engineering.
Offer as a prerequisite
The finding: 70% of traffic had no offer. The decision: evaluate the offer before letting anyone in. No offer, no funnel. The alternative was to keep showing the catalog to everyone and live with dirty metrics. The cost: the flow's traffic would drop by about 70% overnight. It was the hardest decision to defend internally, because it changed how the team measured success.
No plan comparator
The previous flow worked like a store: every package for everyone. The new version shows only the customer's offer. The comparator moved outside the funnel, as a separate discovery surface. You lose in-context comparison. You gain measurable intent: whoever enters, enters to convert.
3-step flow
Evaluation, selection, confirmation. I cut intermediate product-education steps: whoever gets here already passed the offer filter. Every extra screen was an exit point with no function.
100% native
Rebuilding the SPA was the cheap option. Going native cost more development, but it enabled the updated design system, better performance and per-step event tracking. Without that tracking, the new funnel would be as blind as the old one.
THE FLOW
Three steps, no detours
PROTOTYPE
The full flow, in motion
High-fidelity prototype built in Figma, used to validate the flow with product, engineering and business before moving to development.
STATUS
How far it got
The design passed validation with product and engineering and the flow was fully built as a native experience. It didn't reach production before I left the bank.
What was left in place: conversion measured on real offers instead of total traffic, the native flow on the current design system, and per-step event tracking.
Designing a funnel doesn't start at the screen: it starts with the logic of who gets to see what. The decision with the most impact was the one you can't see.
Contact
Let's talk?
I'm open to opportunities as a Product Designer with an AI focus. I'm also interested in collaborations with social organizations and projects that blend technology with purpose. I'm authorized to work across Mercosur and the European Union, both as a freelancer and as an employee.