Generali Arcadia — Securities Master Data Platform

INTRODUCTION

When a financial instrument — a bond, a fund, a structured product — enters an investment firm, someone has to register it. At Generali Investments, that process was split across four different internal tools, with teams re-typing the same data multiple times and no way to see where a request stood. Errors surfaced downstream, too late.

Arcadia was the initiative to fix this — a unified internal platform to govern how financial instrument data is created, validated, and distributed. I joined as Lead Designer to shape the experience end-to-end: research, IA, wireflows, high-fidelity design, and production design system.

The strategic problem

— No single place to start — onboarding split across tools with no shared ownership
— Invisible status — no visibility on where a request stood or who was responsible
— Errors found too late — validation only at end of cycle, when fixes were most costly
— New platform, same habits — Arcadia existed but hadn’t changed how people actually worked

Business trigger — A full business day lost per instrument onboarding. New hires couldn’t learn the process without sitting next to someone for weeks. The cost of doing nothing was growing faster than the cost of rebuilding.

Client

Generali Investments

Date

July 9, 2026

% faster decisions

1

% more data consistency

About the Project

I worked within a large, structured organisation — directly with product owners, tech leads, business analysts, and the Generali Investments MDM operations team.

Design team

— Lead Designer (me) — strategy, research direction, IA, all design decisions
— Collaborating designers — component execution, wireframe iteration, design system build
— Stakeholders — Front Office Asset Managers · MDM Operations · Group Legal Entities · IT Architects

My Responsibilities

Research — Stakeholder interviews, deep-dive sessions, and process analysis. Output: prioritised pain points, data lifecycle map, brief reframe.

UX Lead — Defined overall UX direction. Aligned design decisions with engineering constraints and business priorities across the full engagement.

Mockups — High-fidelity interfaces covering all core flows: instrument onboarding, request management, validation, and status tracking.

Design System — Arcadia DS v1.0: reusable components, tokens, WCAG 2.1 AA compliant. Built for handoff and long-term team scalability.

Context

Research Plan

Research at a glance

— Stakeholder interviews — front office, MDM operators, data analysts; revealed that the core bottleneck was undefined data ownership, not tool complexity
— Process deep dives — live walkthroughs tool by tool; surfaced an average of 4–6 manual re-entry steps per onboarding cycle
— Interaction flows — joint mapping sessions; uncovered dependencies between systems that no one had formally documented
— Wireframe validation — early IA tested iteratively; confirmed users needed status visibility before any UI detail mattered

Key finding: every role had a different mental model of who owned what at each step — and no system enforced a shared one.

Personas

Personas built from research, not assumptions — interviews, workflow walkthroughs, and direct observation of how data actually moves across the Generali ecosystem.

Core tension captured in every persona: front-office speed vs back-office accuracy vs MDM governance. Each profile reflects real behaviours, not idealised ones — pressures, workarounds, and the decision-making patterns that shaped the design.

User Journey Mapping

Journey mapping started from raw interview notes and progressively tightened into structured flows — not the other way around. Early sketches captured actors, steps, and friction points. Later iterations added system states, ownership, and downstream effects.

What the maps revealed

— 3–4 tool switches per task — with no context carried between them
— Invisible handoffs — no one knew where a request was in the process
— Failure at the edges — errors not caught at entry, detected by downstream teams days later

The maps shifted the framing: this was not a UI problem. It was a data lifecycle orchestration problem.

Key Learnings After Research

Users were switching between 3–4 systems per task, re-entering the same data each time, with no visibility on request status or ownership.

This resulted in up to 1 full business day lost per onboarding cycle, errors caught only downstream, and a process that could only be learned by sitting next to someone.

Research: Hypothesis vs Reality

— Initial assumption — better UX layered on the existing Arcadia architecture
— What research revealed — the architecture was the problem: no shared state, no ownership model, no lifecycle visibility across the 4 tools
— Shift in scope — from “redesign the portal” to “design the data governance and orchestration layer”

Old Design — Legacy System

The legacy ecosystem

— Data Hub — master data entry, disconnected from validation tools
— Thin Client — read-only queries, no write access, separate login
— Financial Help Desk — request management via email threads and manual tickets
— Data Extractor — separate export tool, no integration with request status

Each system had different interaction patterns, different access models, and no shared data layer. The screenshots show the fragmentation that made a unified design the only viable path forward.

Wireflows

Wireflows translated the lifecycle map into navigable interaction sequences — tool by tool, step by step.

Key finding: every handoff between systems was invisible. No state was carried from one tool to the next. Designing the handoff moments — what gets passed, who sees it, what triggers the next step — turned out to be as critical as the screens themselves.

— Step-by-step logic across Arcadia Portal, Data Extractor, and Instrument Requests
— Ownership handoffs and validation gates defined at each step
— Edge cases resolved before any visual design began

Wireframes

Wireframes translated validated flows into screen-level structures — hierarchy, action priority, and feedback visibility.

Key finding: users needed to see full lifecycle status at every step, not just their current task. First wireframe iteration showed one step at a time — stakeholder feedback was immediate: “I need to know where this sits in the whole process.” Status became a primary design element, not a secondary one.

— Instrument onboarding — single entry flow replacing multi-tool re-entry
— Request status — real-time tracking replacing email and spreadsheet checks
— Validation — inline feedback at entry, not end-of-cycle

Key Learnings Before the UI

Complexity wasn’t from too many features — it came from hidden dependencies, missing feedback loops, and non-linear workflows with no shared state.

A key pivot

Week two: a senior analyst walked us through their daily process live. Four systems open in sequence. Same identifiers re-typed in each. A shared spreadsheet checked at the end to see if anyone else had touched the record. No system knew what the others were doing. The brief changed on the spot — stop designing screens, start mapping the data lifecycle.

Key design decisions

— Flow Consolidation — one coherent entry path replacing 4 separate portals. Rejected multi-portal model: users kept losing ownership mid-task.
— Dependency Mapping — hidden system dependencies made explicit in the UI. Rejected documentation-only: static docs were outdated within weeks.
— Early Validation — inline error detection at entry, not end of cycle. Rejected end-of-process validation: too costly to fix late.
— Status Visibility — real-time lifecycle tracking per request. Rejected email notifications: no audit trail, no accountability.

High-Fidelity Mockups

ARCADIA Design System v1.0

Arcadia Design System v1.0 — production-ready, WCAG 2.1 AA compliant, built for enterprise data applications.

— Colour: Brand (Primary #971C1E, Navy, Black) + Semantic (Success, Warning, Danger, Info) — all contrast-checked
— Typography: Heading 1–4, Body Long/Short, Caption — optimised for dense data tables
— Components: Buttons (Primary/Secondary/Ghost/Danger × 3 sizes), Form Elements, Badges, Avatars, Cards (4 variants), Data Tables, Icon Buttons, Loading States
— Grid: 12-column, Lucide React iconography
— Handoff: all tokens documented, all states specified — designed for long-term team scalability without design involvement per iteration

Results

Results were validated through moderated usability testing with 7 Generali Investments operators and data analysts, combining timed task sessions with a structured review of the as-is workflow against redesigned prototypes. Post-validation, key metrics were cross-referenced with the Arcadia operations team during a final stakeholder review session.

— +30% reduction in time-on-task for navigation-heavy workflows
Participants completed instrument lookup and dependency-check tasks in significantly fewer steps on the redesigned prototype; average task time dropped from 9.4 min to 6.6 min across the three highest-frequency flows

— +35% improvement in process visibility
Measured via post-task confidence survey (7-point Likert scale): operators reported substantially higher clarity on instrument status and workflow stage after interacting with the new status-tracking model vs the existing dashboard

— −40% reduction in manual reconciliation effort
Assessed through task analysis: the consolidated entry flow and automated dependency mapping eliminated an average of 4.2 manual steps per instrument onboarding cycle, compared to the fragmented multi-screen process in the legacy system

— +35% faster instrument data onboarding
Timed task comparison: end-to-end onboarding of a new instrument (data entry → validation → approval) averaged 12.1 min on the prototype vs 18.4 min on the existing system

Methodology: moderated task-based usability testing (n=7, Generali Investments operators and analysts), think-aloud protocol, timed task comparison, post-task Likert survey, workflow step-count analysis reviewed with the Arcadia operations team.