Reframing a Fire & Life Safety System Around How Operators Reason

RoleLead UX Architect
ClientHoneywell / Notifier
AgencyOgilvy
Timeline13 weeks
Core team~2.5 FTE — UX research, visual design, project management
UX Research & Strategy Interaction Model Design Safety-Critical Systems Information Architecture

Original Hardware Control Panel

Legacy Notifier fire alarm control panel — physical keypad and indicator lights

Redesigned Touchscreen Interface

Redesigned Notifier touchscreen interface — default alarm state with Active Events screen and contextual controls

The redesign moved the platform from a hardware-constrained control panel to a touchscreen product organized around the points operators act on.

Notifier’s fire and life safety control platform had accumulated decades of complexity—device hierarchies, layered configuration rules, and hardware-era navigation—without a current map of its own structure. Its architecture reflected how the system was engineered, not how operators respond to events or navigate under pressure.

I led the UX effort, reconstructing the legacy architecture to make the system legible and working with product and engineering to move beyond the hardware-constrained panel. When comparative concept testing showed that operators oriented around points—not tasks—I used that evidence to recommend a point-centric interaction model for the new touchscreen platform.

Problem

A decades-old fire and life safety control system organized around hardware implementation logic—not how operators respond to events under pressure.

Insight

Comparative concept testing with 16 participants showed that the team’s initial task-centric direction did not match how operators reasoned: they oriented around the affected system point first.

Decision

I recommended reorganizing the product around a point-centric interaction model where point identity, status, and available actions stay together.

Outcome

48% reduction in information-architecture complexity (128 → 66 destinations) while preserving core operational and programming capability.

Scope: Experience architecture, interaction model, research direction, information architecture Partners: Product and engineering

Original Hardware Interface – Key Components

Original Notifier fire alarm hardware control panel — LCD display, status LEDs, soft keys, fixed function buttons, and programming keypad
  • 1

    Status LEDs

  • 2

    Context-shifting soft keys

  • 3

    Fixed function keys

  • 4

    Special function keys

The panel required operators to navigate life-critical functions through a small LCD, fixed hardware controls, soft keys, and a programming keypad. New capabilities had been layered in over time without a clear organizing model.

  • Critical functions were buried in deep navigation paths
  • Operators relied on memorized sequences and changing soft-key behavior
  • The interface reflected implementation logic more than operator decision-making

Before proposing a new architecture, I needed an honest account of what the product already was—and none existed. I built one: clicking through every function and task flow, documenting paths, dependencies, duplicates, and access levels.

Legacy Notifier information architecture — 128 destinations across fragmented technical paths
The legacy information architecture I reconstructed—128 destinations across fragmented technical paths.

Why the legacy structure was difficult to use

Buried individual tasks

Legacy IA detail showing history review task buried beneath multiple branches, filters, date-entry fields, and event-type selection

Common tasks were buried across multiple branches.

Sequential list navigation

Legacy IA detail showing Next and Previous navigation pattern requiring operators to step through related event information sequentially

Function-first entry required operators to navigate to a function, then step through points sequentially using Next/Previous.

From that inventory, I synthesized a functional model organized across a complexity continuum—from event response to expert programming.

Panel

PC

Events
Drill
Tools
Reports
Tests
Configure
Programming
(Panel)
Programming
(PC)
Installing

Simple

Less Time on Task
Little System Knowledge
Light Information Density
Less Access Permissions

User Experience

Complex

Most Time on Task
Expert System Knowledge
Heavy Information Density
Full Access Permissions

The audit revealed a system spanning operational work and expert programming, with duplication and inherited navigation patterns across the continuum—giving design, product, and engineering a shared framework for what to preserve, consolidate, or reframe.

Every navigation decision in the existing interface was downstream of the soft-button paradigm—fixed hardware buttons whose function shifted depending on system state. I pushed the team to test whether that was actually required or just inherited. It was inherited. Working with product and engineering, we moved to a full touchscreen platform—the enabling condition that made both the IA restructuring and the point-centric model viable.

Decision

Shift from a soft-button-constrained hardware product to a full touchscreen platform.

Why

The soft-button paradigm was an inherited constraint from prior hardware generations — not a requirement of the system's operational logic — and it was limiting every downstream design option.

Impact

Moving to touchscreen unlocked persistent navigation, scrollable point lists, and direct selection — the structural properties that made the point-centric interaction model viable.

The team’s initial concept organized the experience around task categories using a home-screen app paradigm. While mapping the domain model, I began to believe a point-based product could be viable—one organized around the specific detector, module, zone, or node requiring attention rather than abstract task groupings.

To test both directions, I collaborated on comparative concept testing using mid-fidelity interactive prototypes of the two proposed interaction models. Testing was conducted in our offices with 16 fire-system installers, operators, and technicians using task-based scenarios and structured walkthroughs.

Method Comparative concept testing with mid-fidelity interactive prototypes of both interaction models
Participants 16 fire-system installers, operators, and technicians
Key Finding Participants consistently thought of and navigated to the affected point before selecting a task—the opposite of the task-first flow the concept assumed
Mid-fidelity wireframe used in comparative concept testing — point-centric Active Events list with expanded alarm row showing inline response steps
One of the mid-fidelity prototypes used in testing—the point-centric model, showing an expanded alarm event with inline response steps.

I recommended reorganizing around the point rather than the task. In the resulting model, the selected point, its status, and its available actions stay together. Instead of navigating to a function and stepping through points with Next/Previous, operators browse a visible list and select the one that needs attention.

Point-centric model wireframe — list-based layout with scrolling rows Point-Centric
Task-centric model wireframe — grid-based app layout with tappable tiles Task-Centric
UX approach Nodes, loops, zones, points of the installed system
UX approach Task categories
Domain structure Managed via context
Domain structure User managed via navigation
Navigation schema Nested / progressive disclosure
Navigation schema Home screen / app paradigm
Primary pattern Scroll to browse, tap to select and expand in context
Primary pattern Navigate to task, complete, return or switch
Decision

Adopt a point-centric interaction model over a task-centric model.

Why

Comparative concept testing showed operators reasoned around affected system points — detectors, modules, zones, and nodes — not abstract task categories, making the task-centric model a structural mismatch for how operators actually work.

Impact

The point-centric model kept point identity, status, and available actions in a single context, reducing the sequential navigation that had required operators to maintain their own mental state across screen changes.

In the point-centric model, the selected point, its status, and its available actions stay together throughout the interaction.

Point list presentation — scrolling list of points within the selected Node, Loop, or Zone

Point list presentation

Scrolling list of points within the selected Node, Loop, or Zone — a contextual set within a region of the system.

Point selection — selecting a point surfaces its status, detail, and available functions

Point selection

Selecting a point surfaces its status, detail, and available functions in context.

Point-specific functions — point detail expanded inline with status, information, and relevant actions

Point-specific functions

Point detail expands inline with status, information, and relevant actions. The task is completed without leaving the list view.

Updated point status — operator returns to updated list with context and orientation preserved

Updated point status

Operator returns to the updated list. Context and orientation are preserved.

The point-centric model converted Next/Previous navigation into direct list selection — point, status, and action remaining together without context switching.

With the point-centric model settled, the architecture had to express it. The largest reductions came from separating routine operational work from specialist programming and consolidating duplicate paths. Event-response logic changed less—safety-critical states were preserved.

Redesigned Notifier information architecture — clearer grouping by recognizable operational domain, reduced depth, and consolidated navigation paths
The redesigned architecture organized the system into recognizable operational domains rather than fragmented technical paths.
128 → 66 48% reduction in information-architecture complexity Retrospective analysis of project artifacts

The reduction reflects a translation from fragmented technical paths into recognizable operational domains—not the removal of necessary capability.

Alarm response was where the point-centric model needed to perform — and where the legacy structure had imposed the most friction.

Before: Legacy alarm response

The legacy workflow required 5–7 actions across multiple screens to acknowledge a single event, with operators maintaining context as they navigated between tasks.

Legacy Notifier interface showing cascading screens an operator navigated during a task-centric alarm response — from system normal through event list, alarm detail, and acknowledgment action

After: Single-context alarm response

The redesigned workflow kept the affected point, its status, and the available action together in one persistent context.

Redesigned Notifier touchscreen interface in default alarm state — Active Events list with prioritized fire alarm and trouble entries, persistent left-side navigation organized by event category
Redesigned Notifier touchscreen interface — expanded alarm event showing inline device detail, location, status, and acknowledge action on one screen
Before
After
Actions to acknowledge 5–7 actions
Actions to acknowledge 1 in-context action
Context shifts 2+ context shifts
Context shifts 0 context shifts
Event visibility 1 event visible at a time
Event visibility Multiple events visible together
Point selection Repeated Next/Previous actions
Point selection 1 direct selection

Action counts are approximate; exact paths varied by event position and whether detail was reviewed.

Redesigned Active Events screen — collapsed view with active alarms surfaced at the top and alarm state visible
The final interface organized system state, event priority, and point-level detail around how operators work — not how the hardware is built.

For the product

48% reduction in information-architecture complexity — 128 destinations consolidated to 66 — while preserving core operational and programming capability.

For operators

Less system structure to learn, with event context and actions kept together throughout response workflows.

For the team

A shared language and architecture — created where none existed — that design, product, and engineering used to discuss, prioritize, and extend the platform.

This project reinforced a pattern I’ve seen across complex-domain work: the first interaction model a team reaches for often mirrors how the system is built, not how its users think. The most consequential design decision here wasn’t any individual screen—it was recognizing that the initial task-centric model was a structural mismatch and having the research evidence to redirect before detailed design locked it in.