Rachael Hunt
← All work

Information architecture for a multi-role product

Structuring a complex product. I designed the navigation for a new multi-role SaaS product, based on card sorts with every user group, and it has held up as the product has grown.

Role
Designer responsible for the information architecture and navigation
Product
Next-gen cloud clinical SaaS, Best Practice Software
Users
Clinicians, admin and finance teams in New Zealand clinics
Methods
Moderated closed card sorts, navigation design

The challenge

A new cloud clinical product had to serve several very different groups, each with their own priorities and access levels, and still feel familiar: the aim was for the product to feel like "home in a new house".

Research with every user group

I identified the main user groups and moderated a closed card sort with each one, to find the themes in how they think about their work and to test our design assumptions.

Card sort results: where clinical users placed five tasks across six categories, with most tasks grouped under Clinical tasks
Card sort results from clinical users: where they expected to find five common tasks.

A structure for every role

From the research I designed a responsive structure and navigation that supports different roles and access levels, with top-level shortcuts for priority workflows: the appointment book, the inbox and patient search.

The product's top navigation and the items in each of its menus: Appointments, Address book, To do, Admin, Finance, Reports, Help, the plus menu and My menu
The navigation and every menu, at a 1440px layout.

The key insight: patients, then action

Every user group's work centred on patients, but patients weren't the end of their workflow: finding a patient was the start of doing something for them. A front-and-centre patient search, with actions right in the results, was a huge win.

Patient search showing recent searches, and results with last and next appointments and quick actions
Patient search: recent searches, last and next appointments, and actions in each result.

Outcome

  • The structure supports 4+ user groups, tailored to different users and clinical types.
  • It fluidly supports priority workflows with top-level shortcuts.
  • It has adapted well and held up as the product has grown.
  • Positive feedback: users said the product felt like "home in a new house".

How I'd use AI now

This project was done before AI tools were part of my process. If I did it today:

  • Use AI to cluster card sort results and open comments alongside the standard analysis, as a second view to challenge my own reading.
  • Simulate how different roles would look for things in a proposed navigation, to find weak labels before a real tree test, which would still decide.
  • Draft navigation labels and help content in several versions for each user group, then test them.