Data Quality management

---

overview

Roles

Lead UX Researcher

Workshop Facilitator

Product Manager

Tools

Miro & Miro AI

Confluence

Jira

Copilot

Company

CVS Health

Duration

3 months,
Oct 24 - Jan 25

---

problem

As CVS Health’s data teams expanded, data owners faced outdated, manual, and fragmented processes for maintaining data quality. They struggled to locate information, identify responsible contacts, and ensure secure handling of sensitive data.

The lack of visibility into lineage and failed checks meant that consumers often discovered issues before owners did. Workarounds for encryption and inconsistent metadata practices introduced both security risks and governance gaps.

Over time, these challenges eroded trust in the enterprise’s data assets, creating a perception that the platform was inefficient and unreliable.

---

goal

Enable business data owners to confidently manage and improve the quality of their data by:

• Identifying the day-to-day experiences, pain points, and unmet needs of data owners

• Establishing clear ownership, lineage visibility, and proactive alerting mechanisms

• Reducing reliance on manual processes and reactive firefighting

• Aligning stakeholders around user-driven priorities through collaborative workshops and research artifacts

---

OUTCOME

This three-month initiative delivered a measurable shift in how data owners managed quality. By grounding priorities in user research and collaborative workshops, the project:

• Reduced resolution times and improved proactive monitoring

• Increased adoption of underused tools through templates and workflows

• Strengthened governance by clarifying ownership and lineage

• Restored trust in enterprise data assets, enabling teams to make faster, more confident decisions

---

user

BUSINESS DATA OWNERS

Common Roles: Business Leaders & Directors

Data Literacy: Medium (on a scale Low-Medium-High)

Pain Points

• No visibility into lineage or failed checks

• Difficulty finding documentation or responsible SMEs

• Reactive firefighting due to a lack of proactive alerts

• Security risks from inconsistent encryption practices

Goals

• Maintain high-quality, relevant data aligned with business needs

• Detect and resolve issues before consumers

• Ensure compliance and security without excessive manual overhead

Behaviors

• Collaborate with technical teams rather than fixing the asset themselves

• Prioritize data quality as a driver of business decision-making

• Balance governance requirements with operational efficiency

---

PROCESS

Discovery & Hypothesis Framing

Activities

1. Facilitated remote workshops with stakeholders using Lean UX Canvas, RACI charts, and Google’s HEART metrics

2. Conducted 10 user interviews and shadowing sessions with business data owners

3. Leveraged AI to draft interview scripts, brainstorm metrics, and consolidate findings

Learning

1. Owners were lacking visibility into lineage and failed checks

2. Documentation was outdated or inaccessible

3. Owners cared deeply about data quality but lacked time and tools to address issues

OUTCOMES

1. Established alignment on user-driven priorities

2. Three hypotheses validated

   • Users lacked real-time tracking of their own assets

   • Lineage visibility had been absent since the downing of our last tool

   • There was a delay in issue detection for owners - their consumers detected them first

IDENTIFIED POINTS FOR IMPROVEMENT

1. Needed clearer standards for metadata and column naming

2. Required proactive alerting to shift owners from reactive to preventive workflows

journey mapping & prioritization

Activities

1. Created proto-journey maps of current-state experiences

2. Facilitated a two-part prioritization workshop using the MoSCoW method

3. Synthesized patterns from sticky-note clustering and AI-assisted theme sorting

Learning

1. Data owners were not the ones fixing data; entry-level engineers and contractors handled most issues

2. Owners were overwhelmed by constant pings and lacked visibility into issues before consumers

3. Security risks emerged from ad-hoc encryption workarounds

OUTCOMES

1. Prioritized features: proactive alerting, lineage visualization, SME routing, and metadata validation

2. Established a shared understanding of user pain points across governance, product, and design

IDENTIFIED POINTS FOR IMPROVEMENT

1. Needed to automate the integration of governance policies directly into workflows

2. We also needed scalable solutions to reduce dependency on manual interventions

Data Quality Management Journey Map - Click to Access board in Miro

---

key Metrics

Objectives

• Improve visibility into lineage, ownership, and data quality issues

• Reduce time-to-resolution for data quality incidents

• Increase adoption of metadata quality tools and proactive monitoring

• Strengthen trust in enterprise data assets

Key Metrics

• Avg. response time to data inquiries reduced from 5 days → 48 hours

• Avg. time to resolve data quality issues reduced from 11 days → 6 days

• % of tables passing quality checks increased from 62% → 82%

• Active usage of metadata quality tools increased from 52% → 62%

• Data owner satisfaction score during this period improved from 1.8 → 3.8 out of 5

---

outputs

• Journey maps of current-state and prioritized future-state experiences

• Stakeholder alignment artifacts: Lean UX Canvas, RACI chart, HEART metrics

• Prioritized backlog of features (proactive alerting, lineage visualization, SME routing)

• AI-assisted synthesis reports and workshop documentation

---

Thank you for your time.