002Honeywell · 2026
Metric Builder
A composable dashboard platform for building management systems
Summary
Led UX end to end for Metric Builder, a composable dashboard platform for Honeywell's Niagara Cloud Suite: from the framing workshop through information architecture, widget flow and the Ask Niagara AI layer.
- Role
- Senior UX Designer
- Client
- Tridium (Honeywell)
- Team
- PM / Architects / Engineering
- Timeline
- End to end, research through architecture
- Platform
- Cloud & Web
My role
I owned UX end to end: research, information architecture, interaction design and the AI layer. I worked closely with product management, solution architects and engineering.
I ran the workshop that reframed the product strategy from customer-specific dashboards to a reusable builder, and advocated for the Asset and Space Mapping restructuring the design depended on.
Impact
94%
valid data connections with smart filtering, vs 67% in the data-first prototype
15–20 min
integrator time to first dashboard, vs 2–3 hours previously
7 in 10
NDS subscribers expressed upgrade interest post-demo
Context
The world behind the walls
Every building you spend time in (an office, a hospital, an airport) has a hidden nervous system. Honeywell's Niagara Framework is the platform connecting it all globally, and Niagara Cloud Suite is its next evolution.
Two players matter here. System Integrators are the builders: contractors who configure and maintain building automation across dozens of sites at once. Facility Managers are the operators: on-site teams responsible for keeping buildings running every day.
How might we help facility teams independently make sense of complex building data to drive faster decisions?
Problem
The data was there. The understanding wasn't.
NCS held telemetry, alarms, energy and asset data across connected buildings. There was no meaningful way to turn any of it into something a user could act on. Finding a single data point meant navigating six steps of device hierarchy.
- 01Limited visualisation capability: NCS relied on flat alarm lists and raw CSV exports.
- 02No reuse mechanism for integrators: every customer view had to be rebuilt from scratch.
- 03Platform-centric, not user-centric data: it did not match how facility managers think about buildings.
Strategy
From a product direction to a design strategy
The product team knew NCS needed visualisation. The open question was how to design it for a cloud-native audience.
A cross-functional workshop surfaced the framing shift that carried the project. The team was focused on customer outcomes, but the people delivering those outcomes had no scalable tool to do it. Solving for the integrators and facility managers would serve the customer far better than solving for the customer directly.
I mapped three competitors alongside this to understand how cloud BMS platforms approach visualisation, and where the gaps were.
A fixed dashboard solves one customer's problem once. A builder solves every customer's problem, and indefinitely.
Research & insights
What the research confirmed, and what it added
Field sessions with facility managers in active buildings, observing workflows in real time rather than in a controlled setting, plus conversations with integrators managing large portfolios.
Nathan, a system integrator across 50+ customer sites: “I'm on a call with a panicking facility manager. I need the answer in two minutes.” He needed reusable views across similar sites, portfolio-level visibility without site-by-site navigation, and professional deliverables without manual effort.
Jean, a facility manager across 8 buildings: “My day starts with alerts. I still can't tell which one actually needs me right now.” She needed one view of building health, context behind an alert without calling an integrator, and role-appropriate information instead of raw technical data.
- 01Users were building their own reporting layer. Because the platform couldn't connect data across systems, facility managers had developed shadow workflows.
- 02The platform spoke device language. Users think in buildings, systems and zones; NCS was organised around devices and controllers.
- 03Integrators were being turned into helpdesk operators. With no portfolio-level health view, the platform forced reactive response instead of proactive monitoring.
- 04Every rebuild erased accumulated knowledge. The time cost was visible; the knowledge cost was not.
I advocated for Asset Mapping and Space Mapping as a foundational prerequisite, restructuring the data hierarchy around equipment and spaces rather than rigid device trees. A significant architectural ask, and the thing that made everything after it possible.
Exploration
Six directions before committing to one
A builder is a category, not a design. I explored multiple paths to uncover what could be absorbed into the final solution.
- 01Spreadsheet model: browse all data in a table, select rows, visualise.
- 02Conversation model: “Show me energy for Building A last 30 days” generates the widget.
- 03Lego block model: pre-configured blocks snapped together.
- 04Map-first model: the canvas opens as a floor plan; click a zone and widgets appear contextually.
- 05Alert-driven model: the builder seeds from active alarms.
- 06Visualisation with smart filtering: the user picks the visualisation first, and the system auto-filters to compatible points. Chosen.
Visualisation-first was the only direction that solved data compatibility, the empty canvas problem, hierarchy navigation and integrator reusability at once. The visualisation choice carries semantic meaning about what data type is needed, which makes it the natural filter trigger.
Solution
Metric Builder in four layers
Layer 01: Template gallery. Templates, not blank canvases. Purpose-built starting points pre-configured with sample metrics show what a meaningful dashboard looks like before any live data is mapped: Portfolio Performance for integrators, Energy & Sustainability, Alarm Intelligence, Software & Firmware Status, Equipment Health, and a blank canvas for expert users.
Layer 02: Widget flow. Two steps, no wrong answers. Step one asks what to show; step two shows only data valid for that choice. The system does the heavy lifting so the user only ever sees options that work.
Layer 03: Canvas & configuration. Depth reveals itself. Every widget configures across two tabs, Data and Display, and the visualisation type can switch while the underlying data stays intact.
Layer 04: Intelligence. “Ask Niagara” lives inside the dashboard. A facility manager asks which building has the most critical alarms; the system surfaces them, explains the context, and lets the user act.
Impact
Early signals from testing and beta
NCS Builder and Ask Niagara are in pre-release, so these numbers come from structured evaluation rather than production telemetry: moderated sessions with think-aloud protocol across integrators and facility managers in multiple customer organisations; an unmoderated beta where selected customers used the builder in real workflows over two to four weeks; and a feedback survey measuring confidence shift, time-to-first-dashboard, data findability and intent to replace existing reporting.
The people closest to the product didn't have a clear picture of what the people using it had to go through. Closing that gap is where real design leadership lives.
Kept high-level to respect confidentiality and data privacy. Happy to talk through the detail offline.