Complex operations do not need another dashboard. They need a clearer path to a decision.
Complex operations do not need another dashboard. They need a clearer path to a decision.
Complex operations do not need another dashboard. They need a clearer path to a decision.
Complex operations do not need another dashboard. They need a clearer path to a decision.
I redesigned fragmented supply-chain reporting into a hierarchy-aware investigation system—helping teams move from the KPI that changed to the driver behind it.
I redesigned fragmented supply-chain reporting into a hierarchy-aware investigation system—helping teams move from the KPI that changed to the driver behind it.
I redesigned fragmented supply-chain reporting into a hierarchy-aware investigation system—helping teams move from the KPI that changed to the driver behind it.
I redesigned fragmented supply-chain reporting into a hierarchy-aware investigation system—helping teams move from the KPI that changed to the driver behind it.
Supply Chain Performance overview — product stage walkthrough from loading to empty heatmap to investigate stale
Duration
1 month
My Role
Lead Product Designer
Team
2 Product Managers
2 BI Developers
2 Data Scientists
3 Engineers
Audience
Inventory planners
Retail operations teams
Category managers
Supplier / brand partners
Scope
Research · UX · UI · Information architecture · KPI structuring · Prototyping · Testing
Tools
Figma
Excel
Highcharts
Jira
Industry
Retail analytics
Supply chain planning
Inventory performance
Platform
Web · Desktop-first enterprise reporting
Users Impacted
Brand managers, promotion analysts, category managers, and supply chain managers across supplier and retailer workflows
Data Coverage
Sales + inventory performance across product hierarchy, store hierarchy, and distribution centers
[02]The pressure
Fragmented reporting hid the driver behind the KPI.
Teams could see that a number moved. They could not reliably walk from that signal to the hierarchy, exception, and operational cause in one path.
01
Scattered sources
Inventory and sales lived across separate reports, so out-of-stocks and store impact required manual stitching before anyone trusted a diagnosis.
Signal risk
02
Hierarchy blind spots
Product, store, business, and distribution breakdowns were hard to keep in context while drilling — slowing QBR prep and exception review.
Investigation cost
03
Slow path to action
Without one trustworthy workflow, teams spent hours reconciling before they could decide what to fix first.
Decision lag
Legacy PrecimaOlder reports — density without a clear exception path · Summary Table
[03]Who needed what decision
Three roles. Three different reasons to trust the same path.
Analysts needed continuity, managers needed defensible evidence, and reviewers needed to know when a precise-looking number deserved caution.
01Analyst signal
Keep the investigation connected.
Move from a KPI to the relevant hierarchy without rebuilding filters or losing the time period.
Design response
Default to exceptions first, then preserve the full inventory path for deeper review.
02Manager signal
Carry evidence into the review.
See the driver, its operational consequence, and a snapshot that can support a decision.
Design response
Add alert history, affected locations, and an exportable review snapshot.
03Data trust signal
Know when the metric needs context.
Delayed or partial store feeds changed how teams should interpret Days on Hand and velocity.
Design response
Put confidence states and calculation context directly beside key metrics.
[04]Success criteria
Translate a dense business brief into observable product behavior.
The requirements were not treated as a feature list. They became a traceable contract between the business question, the interaction, and the evidence the interface needed to preserve.
Business question
UX response
Where are we underperforming?
Surface KPI summary row first
What is the inventory impact?
Unified sales + inventory view
Which child segments are driving it?
Coordinated bottom-table breakdown
Is it product, store, or DC level?
Flexible hierarchy selection model
What should teams investigate next?
Continuation path to trend / export
Requirements mapQuestion → response structure that drove the reporting IA
01
Business question
Where did performance change?
Outcome
Hierarchy entry
02
Hierarchy continuity
Carry filters, scope, and time period through every drill.
Outcome
Exception focus
03
Metric trust
Show freshness and calculation context beside the number.
Outcome
Driver evidence
04
Review handoff
Prepare evidence for the next conversation.
Outcome
Review ready
[05]Design constraints
Simplify the path without flattening the analysis.
The challenge was not a lack of data. It was making the investigation easier while preserving the depth enterprise teams still needed.
Preserve analytical depth from legacy tools while simplifying the scan.
Keep parent/child context visible during drill-down.
Unify sales + inventory so stockout and performance stories meet.
Scale the same reporting logic across enterprise configurations.
[06]Intervention
Four principles that shaped the investigation system.
Design Principles Map — exception-first, continuous context, comparative clarity, and review-ready evidence — with interface behavior proof beside each principle.
01
Exception-first hierarchy
Surface stockout risk, overstock, and velocity anomalies before the full inventory table.
Interface behavior
Exceptions panel prioritizes issues by impact score and confidence.
Top exceptions
Issue
Impact
Status
Stockout risk
98
Investigate
Overstock
86
Review
Velocity drop
72
Monitor
See all (128) →
02
Context continuity
Carry filters and hierarchy through Product → Store → DC drill-down.
Interface behavior
Path bar preserves context and active filters across levels.
PeriodCategoryRegion
Inventory path
Product A→Store 214→DC 03
View path →
03
Comparative clarity
Show prior period, YoY, benchmark, and peer-set comparisons before export.
Interface behavior
Comparison matrix is visible in-line with consistent metrics.
Product A · Store 214
Metric
Current
Prior
Signal
On hand
12.4K
11.8K
↓
Sales
$428K
$401K
↑
Velocity
2.1
2.4
↓
04
Review-ready evidence
Package issue, exposure, locations, confidence, and next action into a review snapshot.
Interface behavior
One-click snapshot compiles the right evidence and recommended action.
Review packet
Issue
Exposure
Locations
Confidence
Next action
Share packet →
[07]Product decisions
Four choices kept the product focused.
The strongest story here is not that every idea worked. It is how scope was reduced without removing the investigation depth teams depended on.
01
Tested
A KPI-first dashboard
Exception queue
Decision
Keep the KPI strip compact and let operational exceptions lead.
02
Tested
Separate root-cause flows
Filter context persistsProduct lens
Decision
Use one persistent drawer that can open from a KPI, heatmap, queue, or row.
03
Tested
Full drill tables on Overview
Comparison matrix
Decision
Keep Overview scannable and move hierarchy depth into Investigate.
04
Tested
Four tabs by default
QBR snapshot
Ready
Decision
Use Overview, Investigate, and Trends as the core. Keep Distribution conditional.
[08]Convergence
Scan → select a lens → drill → validate.
Low-fidelity sketches tested one continuous investigation path before the final tabs were assembled.
01
Scan exceptions
Exception queue
Issue
Impact
Stockout risk · Dairy
98
OTIF drop · Region West
86
Velocity anomaly
72
Overstock signal
64
Full table below
Decision
Surface the highest-risk exceptions before the full inventory table.
02
Select lens
Investigation lens
Dairy→Product view
Decision
Preserve product, store, DC, period, and issue context.
03
Drill into locations
Dairy OTIF72.3%–6.8pp
Affected locations
Location
Variance
Store #1142
–4.2pp
Store #0891
–2.1pp
Store #2204
–0.8pp
Decision
Show the operational driver behind the exception.
04
Validate and export
Review packet
Issue
Exposure
Confidence
Export
Decision
Validate the finding before it becomes review evidence.
This sequence became the interaction contract for Overview, Investigate, and Trends.
[09]Workflow
Reusable modules for the moments that carry decision risk.
Teams scan region × category performance in one view, then focus a cell to investigate without losing surrounding context.
Exception queue
What needs attention now
Exceptions surface by severity so analysts triage stockout and velocity risk before opening the full report.
Drill stack
From summary to root cause
Parent and child tables stay linked as users drill — with clear end-states when the hierarchy cannot go further.
Recent exceptions
Alert ledger with freshness states
Recent exceptions keep severity, path, and data-trust signals visible so review handoffs do not start from a blank export.
[10]Framework
Three views, one continuous investigation.
Overview directs attention. Investigate preserves depth. Trends tests whether the change is temporary or structural.
Final reporting experience — product screens
Exception Overview
See risk instantly
The overview tab became the business entry point — where risk concentrates, which KPIs need attention, and what to review before drill-down or escalation.
Path 01 of 03: See risk instantly
[11]Validation
Validation changed the product—not just the presentation.
Each method was tied to a concrete product decision. The useful evidence is what changed next.
Analyst
Manager
Reviewer
Handoff friction
01Process map
Workflow audit
Mapped where filters, hierarchy, and time period were lost. The redesigned report preserved all three across the investigation path.
Scan
Select lens
Drill
Validate
Preserve hierarchy, filters and period
02Journey map
Investigation journey
Tested the order of attention and drill-down. The interface became one continuous path: scan, select a lens, drill, then validate before export.
LatencyFreshness
CompletenessCoverage
CalculationDefinition
03Data audit
Metric trust audit
Reviewed freshness, calculation context, and ambiguity. Confidence states moved beside metrics that carried review risk.
Hierarchy
Filters
Time period
Driver evidence
Review snapshot
04Evidence log
Escalation evidence
Compared raw exports with what managers needed to defend a conclusion. The final export became a QBR-ready snapshot.
[12]Impact
What the investigation path unlocked.
Four outcomes from the redesigned path — faster prep, surfaced exposure, quicker detection, and strong adoption.
~90 min
QBR prep time-to-insight (from 4+ hrs)
$2.1M
Stockout exposure surfaced
60%
Faster issue detection
85%
Adoption among target users
[13]Next step
Same decision craft for your next operational fire.
Inventory, supply, or exception systems — a focused sprint for an investigation spine, explainable risk, and proof stakeholders can trust.
