Led the redesign of a KPI system that cut time-to-root-cause by 35%, improving triage speed and operational confidence at scale.
Led the redesign of a KPI system that cut time-to-root-cause by 35%, improving triage speed and operational confidence at scale.
The real problem wasn’t reporting - it was reaction speed
The real problem wasn’t reporting - it was reaction speed
Honeywell’s operations teams managed critical assets under strict SLAs, where delayed response directly translated into financial penalties and customer risk.
While KPIs technically existed, they were:
Fragmented across views
Difficult to interpret in context
Slow to act on during live incidents
But the old experience? It didn’t scale.
But the old experience? It didn’t scale.
Noise increased, signals degraded, and root-cause analysis slowed under real-world load.
My Contribution
Critical Assets KPI Testing
Rebuilding the signal layer for faster root-cause decisions
This wasn’t just a UI upgrade. It was a recalibration of how operational signals surfaced under pressure.
Data scientists → breach logic + criticality granularity
Prototyping two concepts (aggregated vs. context-rich KPIs)
Running 12 usability sessions with warehouse engineers & data scientist
Partnering with devs to stress-test tree tables + sparkline feasibility
Result
“Seeing the dip and the timestamp together was a game changer.”
- Sr. Engineer, Ops
Impact: Faster decisions, clearer ownership
By restoring context at the point of failure, the system shifted from passive monitoring to active diagnosis.
Time-to-triage: ↓ 22% (Concept 2 vs. Concept 1)
2× faster fault identification with event
overlays
2.3× more widget interactions than
baseline
This confirmed that clarity, not density, was the unlock for scale.
My Contribution
Critical Assets KPI Testing
Rebuilding the signal layer for faster root-cause decisions
This wasn’t just a UI upgrade. It was a recalibration of how operational signals surfaced under pressure.
Data scientists → breach logic + criticality granularity
Prototyping two concepts (aggregated vs. context-rich KPIs)
Running 12 usability sessions with warehouse engineers & data scientist
Partnering with devs to stress-test tree tables + sparkline feasibility
Result
“Seeing the dip and the timestamp together was a game changer.”
- Sr. Engineer, Ops
Impact: Faster decisions, clearer ownership
By restoring context at the point of failure, the system shifted from passive monitoring to active diagnosis.
Time-to-triage: ↓ 22% (Concept 2 vs. Concept 1)
2× faster fault identification with event
overlays
2.3× more widget interactions than
baseline
This confirmed that clarity, not density, was the unlock for scale.
The fight: Design ambition vs.
Dev feasibility
The fight: Design ambition vs.
Dev feasibility
Design Ideal
Design Ideal
Tree-structured tables for parent → sub-asset clarity
Tree-structured tables for parent → sub-asset clarity
KPI-level sparklines + red-dot threshold markers
KPI-level sparklines + red-dot threshold markers
“My KPIs” with per-user save states
“My KPIs” with per-user save states
Instant deep-dives on click
Instant deep-dives on click
Designed for RCA-first thinking
Designed for RCA-first thinking
Strategy
Strategy
Dev Reality
Dev Reality
Current table component can’t support hierarchical nesting
Current table component can’t support hierarchical nesting
Chart lib can’t render micro-trends with events inline
Chart lib can’t render micro-trends with events inline
High config complexity + backend rework
High config complexity + backend rework
Modal flow required async data fetch + loading UX
Modal flow required async data fetch + loading UX
Built with reporting-first foundation
Built with reporting-first foundation
Constraints
Constraints
Product
Soooooo…..we recalibrated
Soooooo…..we recalibrated
As we explored deeper RCA workflows, it became clear the platform was built on a reporting-first foundation - not real-time diagnostics.
Rather than pushing for brittle, high-maintenance solutions, I worked with engineering to treat constraints as design inputs, not blockers.
Through weekly design-dev syncs, we identified a scalable middle ground:
One-level hierarchical disclosure instead of deep nesting
Progressive reveal patterns to preserve speed without backend rework
Event-aligned sparklines that delivered context without custom charting
Understanding the Frontline: Who monitors what, and why?
Understanding the Frontline: Who monitors what,
and why
Point me to the
next fire to
fight.
Field Service Engineer
Real-time signal, quick RCA path
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Point me to the
next fire to
fight.
Field Service Engineer
Real-time signal, quick RCA path
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Patterns, not just points.
Data Scientist
Trends, anomaly markers, export slices
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Patterns, not just points.
Data Scientist
Trends, anomaly markers, export slices
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
I need to explain this to
leadership.
Maintenance Manager
High-level + drilldown views
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
I need to explain this to
leadership.
Maintenance Manager
High-level + drilldown views
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Features
One framework - multiple insights
One framework - multiple insights
Inline KPI Trends
So… it’s like a heart monitor, but for machines. Instead of staring at a flat “93%,” users could now see the rise and fall of KPIs. Red dots flagged threshold breaches, so they instantly knew when and why things dipped.
Compare Modal
Line up Today, Yesterday, and Target side-by-side - no exporting to Excel, no guesswork. Managers loved it because it turned “uptime slipped” into a story with proof.
My KPIs
Engineers, managers, and DSs each got to save their go-to KPIs without cluttering the global view. Think of it as “My Mix” for dashboards.
Event Overlays
Subtitles for KPI graphs.
True story. Every fault, jam, or anomaly showed up as an overlay on the trendline. Suddenly, that weird 2 am downtime spike made sense - “Oh look, Conveyor 2405 jammed."
From High-Level Summaries to Micro-Insights
From High-Level Summaries to
Micro-Insights
Designing charts, tables, and KPI visualization patterns that scale with understanding.
5%
10%
15%
20%
25%
Scalability
Tree-table exploration for nested KPIs, without frying performance.
Smarter defaults → Role-based presets
Early-warning signal detected
Compare to baseline performance
Highlight abnormal patterns
Predict potential failure window
Recommend preventive task
Fix Triage, Fix Everything
A redesigned workflow that brought focus, speed, and structure to operations.
Adaptive KPI Signals
Highlight what matters most as machine conditions shift in real time.
Unlock 2.3× more engagement through interactive, drill-ready KPI widgets.
5%
10%
15%
20%
25%
Scalability
Tree-table exploration for nested KPIs, without frying performance.
5%
10%
15%
20%
25%
Scalability
Tree-table exploration for nested KPIs, without frying performance.
Smarter defaults → Role-based presets
Smarter defaults → Role-based presets
Early-warning signal detected
Compare to baseline performance
Highlight abnormal patterns
Predict potential failure window
Recommend preventive task
Fix Triage, Fix Everything
A redesigned workflow that brought focus, speed, and structure to operations.
Early-warning signal detected
Compare to baseline performance
Highlight abnormal patterns
Predict potential failure window
Recommend preventive task
Fix Triage, Fix Everything
A redesigned workflow that brought focus, speed, and structure to operations.
Adaptive KPI Signals
Highlight what matters most as machine conditions shift in real time.
Adaptive KPI Signals
Highlight what matters most as machine conditions shift in real time.
Unlock 2.3× more engagement through interactive, drill-ready KPI widgets.
Unlock 2.3× more engagement through interactive, drill-ready KPI widgets.
A/B Testing
A/B Testing
A/B Testing
Validating the Impact of Inline Context
Validating the Impact of Inline Context
Validating the Impact of Inline Context
Rather than assuming clarity, we tested competing ways of surfacing signals - static vs dynamic KPIs, trend-only vs event-aligned views and kept what consistently helped teams act faster under pressure.
Rather than assuming clarity, we tested competing ways of surfacing signals - static vs dynamic KPIs, trend-only vs event-aligned views and kept what consistently helped teams act faster under pressure.
Test A: cards + links
Test B: sparklines + compare modal
The result?
The result?
CLEAN: −32% time to first insight
Concept 1 feels lighter, but I still have to dig
CONTEXT: −41% RCA time, fewer dashboard hops
Concept 2 tells me the story in one glance
CONFIDENCE: Faster decisions, fewer escalations
Now I can explain uptime dips without extra charts
Giving teams back 35% of
their troubleshooting time
Critical-only toggle → decluttered the KPIS
Cuts through the noise, keeps users focused on urgent assets.
Sparklines + thresholds → made trends visible at a glance
Tiny trends that save big headaches. Users instantly see rise/drop patterns.
Compare modal (Today vs Yesterday vs Target) → managers got context in 1 click
One-click RCA: today, yesterday, and target - all in one view.
From 6 clicks to 1: making root
cause analysis actually fast
My KPIs
Custom dashboards without creating chaos. Personal relevance, platform-wide consistency.
Event overlays (red dots)
Faults and KPI dips finally talk to each other.
Export slice
DS-friendly. Engineers stopped asking for raw dumps.
We proved that adding context and smart filtering makes Critical Asset KPIs actionable, but this is far from the finish line. The real challenge? scaling beyond 10 KPIs to 100+ without turning the dashboard into a scrolling nightmare.“If v1 was about clarity, v2 will be about resilience ensuring the system doesn’t just work for 10 assets, but scales elegantly to fleets of 100+.”
We proved that adding context and smart filtering makes Critical Asset KPIs actionable, but this is far from the finish line. The real challenge? scaling beyond 10 KPIs to 100+ without turning the dashboard into a scrolling nightmare.“If v1 was about clarity, v2 will be about resilience ensuring the system doesn’t just work for 10 assets, but scales elegantly to fleets of 100+.”
The real problem wasn’t reporting - it was reaction speed?
Honeywell’s operations teams managed critical assets under strict SLAs, where delayed response directly translated into financial penalties and customer risk.
While KPIs technically existed, they were:
Fragmented across views
Difficult to interpret in context
Slow to act on during live incidents
Led the redesign of a KPI system that cut time-to-root-cause by 35%, improving triage speed and operational confidence at scale.
Led the redesign of a KPI system that cut time-to-root-cause by 35%, improving triage speed and operational confidence at scale.
The fight: Design ambition vs. Dev feasibility
Design Ideal
Tree-structured tables for parent → sub-asset clarity
Tree-structured tables for parent → sub-asset clarity
KPI-level sparklines + red-dot threshold markers
KPI-level sparklines + red-dot threshold markers
“My KPIs” with per-user save states
“My KPIs” with per-user save states
Instant deep-dives on click
Instant deep-dives on click
Designed for RCA-first thinking
Designed for RCA-first thinking
Dev Reality
Current table component can’t support hierarchical nesting
Current table component can’t support hierarchical nesting
Chart lib can’t render micro-trends with events inline
Chart lib can’t render micro-trends with events inline
High config complexity + backend rework
High config complexity + backend rework
Modal flow required async data fetch + loading UX
Modal flow required async data fetch + loading UX
Built with reporting-first foundation
Built with reporting-first foundation
Soooooo…..we recalibrated
As we explored deeper RCA workflows, it became clear the platform was built on a reporting-first foundation — not real-time diagnostics.
Rather than pushing for brittle, high-maintenance solutions, I worked with engineering to treat constraints as design inputs, not blockers.
Through weekly design-dev syncs, we identified a scalable middle ground:
One-level hierarchical disclosure instead of deep nesting
Progressive reveal patterns to preserve speed without backend rework
Event-aligned sparklines that delivered context without custom charting
Understanding the Frontline: Who Monitors What, and Why
Point me to the
next fire to
fight.
Field Service Engineer
Real-time signal, quick RCA path
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Point me to the
next fire to
fight.
Field Service Engineer
Real-time signal, quick RCA path
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Patterns, not just points.
Data Scientist
Trends, anomaly markers, export slices
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Patterns, not just points.
Data Scientist
Trends, anomaly markers, export slices
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
I need to explain this to
leadership.
Maintenance Manager
High-level + drilldown views
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
I need to explain this to
leadership.
Maintenance Manager
High-level + drilldown views
Juan
Verified since April 2019
Trust is the cornerstone of Airbnb's community, and identity verfication is part of how we build it.
Features
One framework - multiple insights
Inline KPI Trends
So… it’s like a heart monitor, but for machines. Instead of staring at a flat “93%,” users could now see the rise and fall of KPIs. Red dots flagged threshold breaches, so they instantly knew when and why things dipped.
Compare Modal
Line up Today, Yesterday, and Target side-by-side - no exporting to Excel, no guesswork. Managers loved it because it turned “uptime slipped” into a story with proof.
My KPIs
Engineers, managers, and DSs each got to save their go-to KPIs without cluttering the global view. Think of it as “My Mix” for dashboards.
Event Overlays
Subtitles for KPI graphs.
True story. Every fault, jam, or anomaly showed up as an overlay on the trendline. Suddenly, that weird 2am downtime spike made sense - “Oh look, Conveyor 2405 jammed."
Giving teams back 35% of
their troubleshooting time
Critical-only toggle → decluttered the dashboard
Cuts through the noise, keeps users focused on urgent assets.
Sparklines + thresholds → made trends
visible at a glance
Tiny trends that save big headaches. Users instantly see rise/drop patterns.
Compare modal (Today vs Yesterday vs Target) → managers got context in 1 click
One-click RCA: today, yesterday, and target - all in one view.
From 6 clicks to 1: making root
cause analysis actually fast
My KPIs
Custom dashboards without creating chaos. Personal relevance, platform-wide consistency.
Event overlays (red dots)
Faults and KPI dips finally talk to each other.
Export slice
DS-friendly. Engineers stopped asking for raw dumps.
We proved that adding context and smart filtering makes Critical Asset KPIs actionable, but this is far from the finish line. The real challenge? scaling beyond 10 KPIs to 100+ without turning the dashboard into a scrolling nightmare.“If v1 was about clarity, v2 will be about resilience ensuring the system doesn’t just work for 10 assets, but scales elegantly to fleets of 100+.”