Automating enterprise risk onboarding to replace manual PDF process

 

Overview

  • My Role: Sole designer creating the automated registration flow

  • Team: 1 Product Manager, 1 Product Owner, 3 engineers

  • Users: External Acquirers, Merchants, and Acquirer Processors


Context

The goal was to automate a manual, PDF-based registration process for High Brand Risk products such as online gambling, skill-based games, and digital currency. There was no existing design system to work from, so patterns were reverse-engineered from the existing site.


The Problem

The legacy process required downloading a PDF form, filling it out offline, and emailing it to a distribution list for the Fraud and Risk Center team to validate and enroll by hand. This created days-long turnaround, a high error rate on incomplete submissions, no way for applicants to track status, and recurring risk compliance delays.

Manual process: download PDF → fill out offline → email to distribution list → FRC team manually validates → FRC team manually enrolls in system

Business impact: days-long turnaround, high error rate on incomplete submissions, poor user experience, no way to track status, operational inefficiency, risk compliance delays


Design Exploration

I explored two directions.

  • Single-page layout — mirrored the original PDF, everything accessible with no navigation, but errors only surfaced at submission.

  • Multi-step progressive flow — used progressive disclosure to show only the relevant fields, surfaced errors immediately, and tested better with users.

The progressive approach won.


Design Solution

I built a three-step registration flow anchored on product type and organization validation.

  • Section headings established clear context

  • Form fields marked required inputs with an asterisk

  • Navigation used a primary "Submit" or "Next" with a secondary "Cancel"

Design decisions

  • Error state combining a red border, icon, message, and ARIA support

  • Two distinct error-guidance paths: an organization that already had an assigned admin (contact relationship manager / contact us) vs. an invalid Acquirer ID (contact relationship manager)

  • A subtle green checkmark for success

  • A "Next" button disabled until the step validated


Cross-Functional Collaboration

  • With Product Management — weekly syncs on priorities; used testing data to justify the multi-step approach and influence scope decisions.

  • With Engineering — partnered to understand technical constraints, participated in sprint planning and QA, and produced detailed component specs covering state variations, animation timing, and error-handling logic.

  • With Stakeholders — presented two design options to the VP of Risk and used user testing data to secure approval and resources.

  • With Users — tested with three Fraud and Risk Center team members in an informal A/B comparison.


What Shipped / Impact

The automated registration system launched as a three-step progressive flow with real-time validation, smart error handling, and clear success confirmation.

  • Registration time dropped from days to roughly four minutes

  • Error rate fell about 70 percent

  • Zero training and little support needed at rollout

  • Component documentation was reused by the team and became the foundation for a broader design system initiative


Reflections

  • What worked: real-time validation prevented errors; demonstrating both approaches side by side built alignment; visual consistency held even without a design system to start from.

  • What I'd change: build in time to create a proper design library; test with external acquirers rather than only internal Fraud and Risk Center staff.

  • What I learned: this project seeded a broader system effort, and documentation is what enables reuse and scale.


Next Case Study

Transforming AWS cloud provisioning workflows through user-centered research and design