Loan Servicing Redesign

Updating the experience where the commercial bankers initiate modifications, renewals and extensions on active loans, from a 13+ step and 4 page workflow to something people can move through.

I led this project end-to-end: defining the problem, running two rounds of research, designing and testing the solution, and handing it off to development. Along the way I used AI to assist with synthesizing and organizing findings that made the research useful beyond our immediate team.

The Starting Point

This didn't start as a redesign project. It started as a question: Does the existing workflow need to be improved, or does it need to be rethought entirely?

The current experience had known usability issues, including too many steps, confusing navigation, no clear way to see what needed action and what was already in progress. However before committing to a direction, I needed to understand whether those were surface-level UI problems or symptoms of a deeper mismatch between how the product was built and how people actually work.

Research

What did the current experience actually require?

Mapped the existing workflow step-by-step.
Documented each step, page transition, and decision point.
Found 17 steps across 4 pages.
1. Filtering before selection.
2. Hidden action controls
3. Confusing user experience

What do we know about loan servicing within the company?

Interviewed internal product specialist who demo and implement the product.
Internal users found the process functional but manual and saw the potential to automate the experience.
We needed to talk to real users to understand the day to day pain points.

How do users work with loan servicing?

Conducted 18 usability sessions across 3 iterative rounds with participants from 9 different financial institutions.
Had built in workarounds so internal team members didn't realize the full picture of their pain points.
Participants rate the new design easier and more efficient.

AI throughout the process

Synthesize notes across sessions.
Surface patterns.
Structure documentation.
capture reasoning behind decisions so the work could be built upon by the broader team.

Research that lives only in one person's head isn't that useful. Make the findings discoverable and traceable.

Defining the Problem

After the research, the problem wasn't that the UI is confusing. It was that the experience was built

The Design

Two directions explored:

Design A

Inline drop-downs on each user row.
Users set the action type per loan before processing.
Familiar, lower learning curve, everything visible and interactive at once.

Design B

Checkboxes with shared action menu.
Select loans, assign action, process.
Faster for bulk operations.

Both directions eliminated the multi-page navigation entirely. The core structure included 2 tables, separating in-progress from credit actionable loans, which gave users an immediate, clear picture of where things stood without having to dig for it.

Across three rounds of testing, neither design emerged as the clear winner. Both were meaningful improvements. Preferences came down to familiarity and workflow habits. Throughout the testing, we were able to eliminate unnecessary or confusing interactions including a summary card and adding to the design systems library to include and define component interactions with drop-downs within tables, bulk table interactions and sticky footer behavior.

Outcomes

Every participant rated the redesign easier and more efficient than their current system.

Participants estimated 70-90% times savings on larger workloads.

Reduced from 4 pages to a single page by eliminating the second page of navigation entirely.

Development scoped and built.

Next Steps

During my user interviews, I asked questions about other processes including reviews and short term extensions. This is another step in loan servicing that needs to be connected to the new process.