Redesigning a treasury payments platform around the people who use it

When I joined Treasury Intelligence Solutions (TIS) as Lead Product Designer, the company was on its third attempt to modernise its platform. Sales were declining, clients were leaving and support tickets kept piling up in Jira. When we looked closer, most of these tickets were not real technical issues. They were caused by unclear workflows and poor UX.

I was responsible for the full redesign of two TIS products, PayOptix and RiskOptix. This case study is about PayOptix, the payments hub of the TIS platform.

In six months we delivered the MVP of the new PayOptix. Eight weeks after the first launch, support tickets dropped by 67%, and in less than a year NPS grew from 4.5 to 8.

• 67% fewer support tickets, 8 weeks after the first launch
• NPS from 4.5 to 8 in less than a year
• 72% of users satisfied with the new design and UX

01 · Analyse Matomo audit to find where users struggle: flows, heatmaps and behaviour analytics.

02 · Listen 15 structured interviews with users of different experience levels.

03 · Define Two new personas based on the people who actually use the system every day.

04 · Redesign Reusable patterns, a scalable design system, prototypes validated with product owners, and new master data, payments, statements, transactions, search and navigation.

05 · Launch MVP in six months, followed by a clear drop in support tickets and a higher NPS.

TIS had around 60 clients. In the previous three years, 10 of them stopped using the platform completely. Another five or six wanted to leave, but their processes were so tied to the solution that migrating was too expensive for them at that moment. The last NPS survey scored 4.5 out of 10.

Our clients were very clear about what was wrong:

• Outdated look and feel and a system that felt too complex
• Broken authorizations: user roles, rights and data were mixed together. Some users couldn't see or use features they needed, while others could see data and pages they should never have access to.
• Confusing navigation, which made it hard to find the main functionality
• Slow filters: the main filter was a tree view that triggered huge database queries just to show a single bank account. The endpoint took about a minute to load, and users were left looking at a blank white screen.

When I joined, TIS had no research practice — just five or six unscoped interviews done for the admin panel, run by a designer with no research background. I put together a user research strategy to find the causes of the high abandonment and support-ticket volume, and presented it to the Director of Product.

Matomo: finding where users struggle

I mentored our designer on using Matomo for flow analysis, heatmaps and behaviour analytics, combined with short, scenario-based interviews to validate hypotheses. This surfaced the three worst-performing areas — Statement Monitor, Payment Monitor and Master Data — giving us a clear starting point instead of auditing the whole platform.

Interviews: talking to the right people

I interviewed 15 users at different experience levels, and found TIS had been recruiting the wrong participants for years — Heads of Treasury and Treasury Managers, not the people using the system daily. From this I built two new personas, Payment Specialist and Statement Specialist (senior and beginner), and every design decision from then on was made for them.

Two findings stood out: chaotic, fragmented master data spread across separate screens repeating the same information, and overloaded flows — payments, statements and transactions crammed into one screen mixing roles, rights and data.

TIS already had a component library. It was a good starting point, but it was more of a prototype than a real design system.

I did comprehensive research to build reusable patterns. To make sure the new patterns met both business requirements and user needs, I created vibe-coded prototypes and validated them together with our product owners.

Building on the existing library, I turned it into a real, scalable design system, compatible with React MUI X, so our front-end developers could implement the designs consistently.

I documented the design system for two different audiences:

• Confluence for the product management department
• Storybook for the front-end team

One Master Data screen

I replaced the separate screens for banks, bank accounts, legal entities, company and account status with one Master Data screen, to reduce the cognitive load and simplify the flow.

The new Master Data screen: banks, bank accounts and legal entities in one grouped, searchable view.

Separate screens for payments, statements and transactions

I split the overloaded screen into separate screens for Payments, Statements and Transactions, one per domain, using feature flags to control access rather than rebuilding the authorization system from scratch.

The Payments screen on its own, with its own approval queue and history.

Tag-based search instead of the tree view

The slow tree-view filter was completely removed. On the Transactions screen, users now search with editable, reusable tags, for example "Bank IS [selection]", instead.

A clear navigation

The new navigation came out of a dedicated research process: Matomo data analysis, a moderated card sorting session using a think-aloud protocol, and a debrief with follow-up interviews to validate the findings. The strongest result was repetitive screens showing the same information from several different entry points — the main driver of the confusing user flows.

• Dashboard
• Banking: Master data, Workflows, Requests, Approvers
• Pay: Payments, Statements, Reports, Transaction discovery
• Fraud detection

The Executive dashboard, built around the widgets specialists check first.

Drilling into a dashboard widget opens the full report, like this 30-day cash activity view.

• Support tickets dropped by 67% eight weeks after the first launch. This showed that the redesign solved the real UX problems behind the tickets.
• NPS grew from 4.5 to 8 in less than a year after the redesign.
• 72% of users were satisfied with the new design and UX, based on a survey about the redesign.

One of my biggest learnings from this project was that good research starts with identifying the right audience, not simply talking to customers.

The existing research had been conducted with Treasury Managers and Heads of Treasury, who influenced purchasing decisions but were not the people using the product on a daily basis. By questioning this assumption, I identified a critical gap in our understanding of user needs and convinced the Director of Product to rethink the research strategy entirely.

This experience reinforced that meaningful insights come from combining the right participants with the right methods. Analytics through Matomo helped uncover where users were struggling, while card sorting and interviews revealed why those issues existed. Together, these methods provided the evidence needed to make confident product decisions.

The most important lesson was that successful redesigns are rarely about visual changes. The measurable outcomes, including reduced support tickets and increased NPS, came from simplifying information architecture, workflows and search based on validated user insights. By focusing on the right users and uncovering root causes rather than symptoms, we were able to deliver improvements that had a tangible impact on both customer experience and business outcomes.