Danske Bank — Research Hub

Start with the reader’s needs, define the right problem, explore alternatives, then validate the experience.

Reconstructed UX framework for this portfolio story. The activities below describe a research and validation plan, not verified sessions or original research artifacts.

1. 01 · Explore broadly

   Discover

   Listen to readers, audit the research catalogue and map discovery, reading and subscription journeys.

   Research questions · journey map

2. 02 · Focus the problem

   Define

   Synthesize observations into a clear problem, navigation priorities and explicit criteria for successful tasks.

   Problem statement · test criteria

3. 03 · Explore alternatives

   Develop

   Compare low-fidelity options for search, category navigation, article hierarchy and subscription selection.

   Wireframes · testable prototype

4. 04 · Validate and refine

   Deliver

   Test the prototype with representative readers, prioritize observed issues and review accessibility before handoff.

   Validation log · refined interface

A central place for Danske's research community to publish, browse, and trace the lineage of every study informing a product decision. Built to be skim-able for execs and rigorous for researchers.

Team: 2 designers · 4 engineers · 1 researcher

Research was scattered across Confluence, drives, and inboxes. Findings repeated themselves because nobody could find what already existed. We needed a hub with shape, not a folder.

How might teams find a relevant study quickly and trace a finding back to its evidence before making a product decision?

1. 01

   Interviewed 18 researchers and 7 PMs to understand how findings actually travel through the org.

2. 02

   Modeled research as a graph — studies, insights, decisions — instead of a flat list.

3. 03

   Designed for two audiences in one shell: scanners and readers.

4. 04

   Established editorial guardrails so the hub stayed scannable as it scaled.

Search is the front door

Search logs showed nobody browsed — everyone searched. Retrieval became the primary architecture, with filters for topic, date, and team instead of a folder tree.

Research as a graph, not a folder

Studies, insights, and decisions were modelled as linked nodes. Every finding carries its lineage, so a claim can always be traced back to its evidence.

Two audiences, one shell

Executives skim 30-second summaries while researchers read full reports. One product serves both through layered depth instead of splitting into two tools.

Editorial guardrails

Templates and publishing standards keep the hub scannable as it scales — structure is enforced by the system, not by discipline.

Conceptual illustrations, not product screenshots or research artifacts.

The hardest part of this project was not the interface — it was agreeing on what a "finding" is. Modelling research as a connected graph forced every team to confront how their work actually travels through the organisation, and that shared model did more for reuse than any screen we shipped.

If I ran it again, I would bring engineering into the taxonomy workshops earlier. The domain model we designed in week six was almost exactly the one we shipped — the weeks in between were spent translating intent that could have been co-authored from day one.