← Back to work
Data Quality Center
One platform that gives stewards, site managers and executives the data-quality view each of them actually needs.
Transforming enterprise data governance through intuitive design. A Fortune 500 manufacturer was managing data quality across 200+ fragmented, technical tools that non-technical stakeholders couldn’t use — I led UX/UI design end to end to bring it under one roof.
65%
Faster issue resolution, from 14 days down to 5
$1.8M
Annual operational savings in the first year
92%
Post-launch user satisfaction (n=45)
CLIENT
Fortune 500 manufacturer
TIMELINE
27 months, 2026
PRODUCT TYPE
Enterprise SaaS · B2B · Data governance
PLATFORM
Enterprise web app
ROLE
Lead UX/UI Designer
TEAM
2 designers, 5 engineers
SCOPE
Research, design system, iteration
STATUS
Launched · post-launch survey n=45
THE PROBLEM
Unmanageable, unusable data quality
200+ fragmented, technical tools made data quality unmanageable — delaying 40% of business decisions and costing $2.5M a year in manual checks.
THE SOLUTION
One platform, role-based views
A unified platform — dashboard, task tracker, quality heatmap, and pipeline monitor — with role-based views for stewards, managers, and executives.
01 — DESIGN PROCESS
A structured process, run iteratively.
We frequently cycled back through phases based on user feedback and testing results, rather than moving through them once.
01
Discover
Research & analysis
User interviews · Contextual inquiry · Competitive analysis · Stakeholder workshops
02
Define
Synthesis & strategy
Personas · User journeys · Problem framing · Success metrics
03
Ideate
Concept generation
Design studio · Sketching · Crazy 8s · Concept validation
04
Design
Interface design
Wireframes · Hi-fi mockups · Prototypes · Design system
05
Test
Validation
Usability testing · A/B testing · Accessibility audit · Iterations
06
Deliver
Implementation
Dev handoff · Design QA · Launch support · Analytics
02 — THE CHALLENGE
A Fortune 500 manufacturer was flying blind.
Data quality was managed across 200+ sources with fragmented, engineering-first tools. Site managers and executives had no clear way to understand issues, assign ownership, or act before problems hit operations.
40%
Of business decisions delayed due to unreliable data
$2.5M
Annual cost from manual data quality checks
14 days
Average time to identify and fix a data quality issue
01
Lack of Visibility
Data stewards had no centralized view of data quality metrics across systems.
02
Complex Workflows
Identifying and tracking data quality issues required navigating multiple tools.
03
Poor Collaboration
No clear communication channel between site managers and data teams.
04
Limited Insights
Executives lacked high-level dashboards to understand organizational data health.
Create unified dashboard
Streamline task management
Enable real-time collaboration
CONSTRAINTS
A fixed 6-month engagement across 200+ live data sources, for a client whose stakeholders spanned three very different mental models — steward, site manager, and VP. Early design decisions had to hold up across all three without forking into separate builds, which is exactly the tension the role-based dashboard decision below was built to resolve.
03 — RESEARCH
Understanding the people behind the data.
Five weeks of mixed-method research across three roles and twelve manufacturing sites grounded the product in real operational needs.
18
USER INTERVIEWS
In-depth interviews with data stewards, site managers, and executives.
87
SURVEY RESPONSES
Quantitative research across 12 manufacturing sites.
8
CONTEXTUAL INQUIRY
Sessions observing users in their natural work environment.
6
COMPETITOR PLATFORMS
Existing data quality tools evaluated in the market.
WHAT WE HEARD
01 — Users spent roughly 40% of their time manually checking data across disconnected systems.
02 — 85% of participants found existing tools too complex to use confidently.
03 — The recurring workflow was: identify issue → assign team → track resolution → verify fix.
04 — Executives needed trend-level visibility, while stewards needed technical depth.
05 — Real-time alerts were essential to prevent issues from cascading unnoticed.
Identify issue → Assign team → Track resolution → Verify fix
KEY INSIGHT
Executives needed trend-level visibility while stewards needed technical depth — one view could never serve both, so the product had to be role-aware from the start.
DESIGN METHODOLOGY
A framework built for enterprise complexity.
With three stakeholder groups holding three different mental models of the same problem, the biggest risk wasn’t bad visual design — it was solving the wrong problem well. The process borrowed its shape from the Double Diamond’s divergent-then-convergent structure and Human-Centered Design’s insistence on continuous user feedback, rather than a single upfront requirements phase.
WHY THIS STRUCTURE
A linear, waterfall-style process would have locked in assumptions before any of the three personas below had a chance to challenge them. Instead, discovery was deliberately kept wide — interviews, contextual inquiry, and competitive analysis all running in parallel — before converging on a single problem statement. The same divergent-converge rhythm repeated at the design stage: multiple dashboard concepts were explored before the role-based model (see the trade-off in Solution) was settled on.
That structure is also why research and testing never fully stopped. Human-Centered Design treats validation as continuous rather than a single gate before launch — which is why usability testing, A/B testing, and accessibility review all reappear inside the Test phase of the process above, rather than happening once at the end.
01
Stakeholder Interviews
Aligned engineering, product, and client sponsors on scope and success metrics before user research began.
02
User Interviews
18 in-depth conversations surfaced the emotional and behavioral context behind each role’s relationship to data quality — not just their stated needs.
03
Contextual Inquiry
8 sessions observing stewards and managers in their actual workflow surfaced friction a survey alone would have missed.
04
Competitive Audit
Evaluated 6 existing data-quality platforms to understand where the market was already failing non-technical users.
05
Heuristic Evaluation
Benchmarked the client’s legacy tools against usability principles, establishing a real baseline to design against and later measure improvement.
06
Card Sorting
Validated how each persona expected data domains and metrics to be grouped, before the dashboard’s information architecture was locked in.
07
Usability Testing
Every major feature — dashboard, tracker, heatmap, pipeline monitor — was tested against real tasks before release, not just reviewed visually.
08
A/B Testing
Compared card-based against list-based dashboard layouts to settle the scanability question with behavioral data rather than opinion.
09
Accessibility Audit
Verified that color-coded quality scores remained distinguishable for colorblind users — critical given how central color-coding was to the design.
04 — USER PERSONAS
Three roles. One shared data model.
Read top to bottom: goals → pain points → core needs. Behavioral archetypes and tooling sit underneath as the evidence bridge.
SC · AGE 34
Sarah Chen
Data Steward
Data Governance Lead | Quality Analyst | Systems Coordinator
Represents 7 of 18 interview participants
“I need one place where I can see everything that’s wrong with our data and who’s fixing it.”
GOALS
Monitor data quality across multiple sources • Quickly identify and resolve issues • Track team performance metrics
PAIN POINTS
Too many tools to manage • Difficult to prioritize issues • Poor visibility into resolution status
CORE NEEDS
Locate the source and owner of an issue quickly • See lineage across systems without switching tools • Prioritize by business impact, not just volume • Verify a fix with confidence before closing it out
BEHAVIORAL ARCHETYPES
Primary: Triage Lead · Secondary: System Navigator
TOOLS
Excel · Power BI · ServiceNow · Snowflake · Collibra
Findings: KI01–03 · Pain Points: PP01–02 · Needs: N01–03
MR · AGE 42
Mike Rodriguez
Site Manager
Plant Operations Manager | Quality Liaison | Reporting Owner
Represents 6 of 18 interview participants
“I shouldn’t find out about data quality issues from my VP. I need proactive alerts.”
GOALS
Ensure site data accuracy • Minimize manual data validation • Report quality metrics to leadership
PAIN POINTS
Unexpected issues disrupt operations • Limited insight into pipeline health • Manual reporting is time-consuming
CORE NEEDS
Get proactive alerts before issues reach leadership • See site-level data health without needing engineering support • Understand what an issue means operationally, not just technically • Generate a leadership-ready report in minutes, not days
BEHAVIORAL ARCHETYPES
Primary: Early-Warning Seeker · Secondary: Operational Translator
TOOLS
Excel · Power BI · SAP · ServiceNow · Teams
Findings: KI04–05 · Pain Points: PP03 · Needs: N02, N04–05
JP · AGE 48
Jennifer Park
VP of Operations
Executive Sponsor | Budget Owner | Board Reporting
Represents 5 of 18 interview participants
“Show me where we stand and where we’re trending. I don’t need the technical details.”
GOALS
Make data-driven decisions confidently • Understand organizational data health • Reduce operational costs
PAIN POINTS
Lack of visibility into quality trends • Decisions delayed by data uncertainty • Can’t quantify ROI of quality initiatives
CORE NEEDS
See trend, not transaction — one health score over time • Trust the number without auditing how it was calculated • Compare data health across sites and business units • Tie data quality investment to a measurable outcome
BEHAVIORAL ARCHETYPES
Primary: Trend Reader · Secondary: Decision Anchor
TOOLS
Power BI · Tableau · Excel
Findings: KI04 · Pain Points: PP04 · Needs: N01, N04
05 — FINAL DESIGN
One platform. Four connected experiences.
01
Unified Dashboard
PROBLEM
Users needed to switch between 5 different tools to get a complete picture of data quality.
DESIGN DECISIONS
Card-based layout for scanability • Color-coded quality scores for instant recognition • Critical alerts prioritized above-the-fold
SOLUTION
A comprehensive dashboard with real-time metrics, trend visualization, and quick actions.
85% reduction in time spent gathering information
02
Task Tracker
PROBLEM
Data quality issues were tracked in email threads, leading to lost context and missed resolutions.
DESIGN DECISIONS
Kanban-style board for visual workflow management • Automated task creation from rule violations • Rich commenting and attachment support
SOLUTION
An integrated task management system with assignment, prioritization, and status tracking.
92% of tasks resolved within SLA, up from 45%
03
Quality Heatmap
PROBLEM
Executives couldn’t quickly identify which data domains or sources had quality issues.
DESIGN DECISIONS
Hierarchical drill-down from org level to individual tables • Tooltip overlays for detailed metrics on hover • Temporal slider to view quality trends over time
SOLUTION
An interactive heatmap showing quality scores across all data assets and business units.
60% reduction in incident rate through proactive management
04
Data Pipeline Monitor
PROBLEM
Data pipeline failures often went unnoticed until they impacted downstream reports.
DESIGN DECISIONS
Node-based visualization of data flow • Color-coded status indicators with error details • Historical performance charts for each pipeline
SOLUTION
A real-time monitoring interface with automated alerting and root-cause analysis.
Mean time to detection: 6 hours → 15 minutes
A TRADE-OFF I’D OWN
Testing showed one shared dashboard didn’t suit every role — a steward needed row-level detail, a VP needed a trend line, and forcing both into the same view satisfied neither. The solution became role-based views on a single shared data model, so every persona sees the right depth without the underlying numbers ever diverging. Role-based dashboards resolved that core tension, but with hindsight I’d surface that fork earlier — before high-fidelity design — rather than discovering it in usability testing.
06 — ACHIEVEMENT
Designing for measurable operational change.
65%
Faster issue resolution — 14 days to 5
80%
Reduction in manual data quality checks
92%
User satisfaction, post-launch survey (n=45)
$1.8M
Annual operational savings, first year
→ Unified platform reduced tool-switching from 5 apps to 1.
→ Automated alerts decreased missed issues by 85%.
→ Executive dashboard enabled data-driven decision making.
→ Design system accelerated feature development by 40%.
The Data Quality Center transformed how the organization manages data quality. By making complex data governance accessible to all stakeholders, we empowered teams to proactively maintain data integrity — accelerating business decisions and reducing costly errors.
07 — REFLECTION
What I’d carry forward.
What I learned
Progressive disclosure has to be structural, not cosmetic. A larger font doesn’t make a metric executive-friendly if the underlying data model can’t support drill-down.
What I’d test next
I’d instrument persona-level product analytics from day one, rather than relying on aggregate satisfaction data alone.