Georgia-Pacific — Six steps, five tools, one screen
Five disconnected monitoring tools consolidated into one failure-mode platform, with hierarchical KPIs surfacing the risks that matter first.
Five disconnected monitoring tools consolidated into a single failure-mode platform.
[ The problem ]
Georgia-Pacific's engineers monitored equipment through a patchwork of third-party tools — spreadsheets, analytics platforms, PI diagrams, alert dashboards — each with its own interface and data structure. Diagnosing a failing digester meant moving through six steps across five of them. The data existed. The path through it did not. Fragmentation slowed troubleshooting, created blind spots, and pulled attention from the problem to the navigation.
[ Goals ]
Consolidation — one platform instead of five disconnected tools. Failure modes first — organised around how equipment actually fails, not how data is stored. Hierarchy — KPIs that surface the risks that matter before the ones that do not. Adaptability — a framework flexible enough for different mills and use cases.
Approach
The work started in a room with plant engineers and a whiteboard, not with a redesign of the existing dashboards. Stakeholder workshops with engineers, analysts and managers mapped the real diagnostic path — six steps across five tools — and agreed what an engineer needs to see first. From there: rapid low-fidelity wireframes in Figma focused on hierarchy and navigation, iterated into clickable prototypes and tested with the same engineers, then a high-fidelity system — reusable components, consistent dashboards, hierarchical KPI views — built to adapt across mills rather than fit one.
[ Scope and constraints ]
[ The work ]
The surfaces.
The path as it was
Understand exactly what an engineer had to do to diagnose one failing asset before designing anything.
Mapped the legacy sequence with the engineers who walked it every day, in workshops and on whiteboards, before opening Figma.
Six steps across five tools — dashboards, asset lists, spreadsheets, PI diagrams, analytics overlays — became the brief in one sentence: one screen.
Equipment overview
Show a plant engineer what needs attention before they go looking for it.
Organised the platform around failure modes rather than data sources, with hierarchical KPIs that surface the highest-risk equipment first.
The first screen answers the question that used to take five tools.
Controller detail
Let an engineer diagnose a specific asset without leaving the platform.
Brought the readings that actually decide a diagnosis — kappa level, brownstock washing, controller performance — into a single view with the history alongside them, so an engineer can move from the overview to a specific asset without changing tools.
The six-step diagnostic collapses into one screen.
Alerts and thresholds
Make the platform tell engineers when something is wrong instead of waiting to be asked.
Designed threshold configuration and failure-mode risk views around the way the platform is organised, so an alert names the failure mode and its associated models rather than a raw metric.
Monitoring shifts from something you check to something that reaches you.
[ The system underneath ]
[ Results ]
5 → 1 | Monitoring tools consolidated 6 → 1 | Steps to diagnose one asset 1 | Design system, built to travel across mills
The diagnostic that used to take six steps across five tools became one screen. Engineers can see what needs attention before they go looking for it, move from the overview to a specific asset without changing tools, and be told when something is wrong instead of checking. Early feedback centred on speed and adaptability: faster troubleshooting, better visibility into critical risk, and a design foundation that other mills can adopt without starting over.
What I carry forward
Rapid iteration with engineers under real deadlines taught me to prioritise ruthlessly: the first screen has to answer the first question. Everything else can be one click away, as long as that click is obvious.