TL;DR
A data governance initiative improves data quality by making a named team responsible for six things: accuracy, completeness, consistency, timeliness, validity and uniqueness.
Most companies do not lack data. They lack agreement. Finance, sales and operations each keep their own version of a customer, a product or a month-end number. Each department fixes its own problems, so the fixes do not match.
Departmental clean-ups and one-off tools fall short because nobody owns the result across the business. The same error is fixed three times in three systems, and it comes back next quarter.
A governance initiative fixes this with three things: people who own the data (from an executive sponsor down to day-to-day data stewards), a repeatable improvement method such as Six Sigma, and tools that map, profile, clean and monitor data.
The payoff is not tidier tables. It is better visibility for customers and partners, easier regulatory compliance, smoother integration after growth or a merger, and knowledge workers who spend their time using data instead of arguing about it.
“Data quality is not an IT task. It is a decision about who is accountable when the numbers disagree. ”
The Problem: Why Data Quality Breaks as Companies Grow
Picture a board meeting. The CFO presents quarterly revenue from the ERP. The sales head presents a different number from the CRM. The COO asks a simple question: “Which one is right, and why do we have two?”
Nobody in the room can answer quickly. The gap is not fraud and it is not a bug. It is a governance problem. Each system was built for its own department, with its own definition of a “closed deal” and its own idea of when revenue counts.
This pattern repeats in almost every growing company:
- Too many systems: Each department picks its own software. Over the years the business ends up with a CRM, an ERP, spreadsheets and point tools that hold overlapping data and do not talk to each other.
- Local fixes: When a department finds bad data, it builds its own clean-up process. Other departments do the same. The fixes overlap, contradict each other and repeat the same work.
- Blocked access: Analysts and managers wait days for an extract from another team’s system, and the file they finally get has gaps, old values or duplicate rows.
The cost is real. Gartner says poor data quality costs organizations at least US$12.9 million a year on average, according to its research from 2020 [1]. Treat that as a directional signal rather than a benchmark for your company. Gartner also reports that 59% of organizations do not measure data quality at all [1], which is part of the problem.
If you lead finance, operations, sales or IT, the question is not whether your data has quality issues. It does. The question is whether anyone is accountable for fixing them across the whole business.
The Four Objectives of a Data Governance Initiative
Companies rarely start governance for its own sake. They start it to reach a business goal. Most initiatives aim at one or more of four objectives: better visibility for internal and external customers, compliance with regulation, smoother operations after rapid growth or a merger, and more efficient knowledge workers. Each is explained below, with independent sources.
1. Better Visibility for Internal and External Customers
When a company shares data with customers, suppliers or distributors, every mistake travels. A wrong weight, pack size or product code in a supplier’s catalogue ends up in the retailer’s system, and from there in shelf labels, delivery plans and invoices. In supply chain management, a buyer, a planner and a supplier all need the same picture of orders, stock and lead times.
A 2017 guide from the USAID Global Health Supply Chain Program on applying GS1 best practices warns: “Without a robust business process in place, the quality of data can deteriorate as it flows through the supply chain” [2]. It describes data governance programs as managing “the actions, methods, timing, and responsibilities for supporting master data”, and notes that “A solid data governance program formalizes accountability for data management across the organization” [2].
Both sides benefit. GS1, the organization behind global barcode standards, says that “By improving the quality of data, trading partners reduce costs, improve productivity and accelerate speed to market” [3]. The USAID guide gives a simple test of whether shared data is good: “electronic data exchanged equals physical data” [2]. In other words, the data about a product should match the product in the box.
2. Compliance with Regulation
Many laws do not just say what data a company may hold. They also require the company to keep that data accurate and to prove that it is under control. Governance supplies that proof: named owners, written rules, lineage and records of who changed what.
Privacy law. Under the EU’s General Data Protection Regulation (GDPR), Ireland’s Data Protection Commission explains that “Controllers must ensure that personal data are accurate and, where necessary, kept up to date”, and that under the accountability principle “the controller is responsible for, and must be able to demonstrate, their compliance” [4]. Without clear ownership of customer data, both are hard to show.
Banking. After the global financial crisis, the Basel Committee on Banking Supervision found that many banks “were unable to aggregate risk exposures and identify concentrations fully, quickly and accurately”. In January 2013 it published its Principles for effective risk data aggregation and risk reporting, known as BCBS 239, to address this [5]. Meeting them depends on defined data owners, consistent definitions across systems and controls on how risk data is put together.
Other laws follow the same pattern, such as India’s Digital Personal Data Protection Act 2023, HIPAA for US health data and SOX for US financial reporting. Rules and deadlines change, so confirm current requirements with your legal team.
3. Improving Operations After Rapid Growth or a Merger
Fast growth adds systems faster than anyone can align them. A merger brings two customer lists, two product catalogues, two charts of accounts and two ideas of what “active customer” or “revenue” means.
A 2026 TechTarget article on mergers and acquisitions notes that merging firms often find it “challenging to integrate incompatible IT systems and consolidate disparate data”, and that weak governance leads to “data siloes that affect cross-functional collaboration and communication” [6]. The open questions it lists are governance questions: how data should be reconciled across platforms, which data model becomes the standard, and who owns which data. Its recommendation is “clear system ownership and strong accountability”, with “unambiguous decision rights and authority” [6].
4. Making Knowledge Workers More Efficient
A McKinsey Global Institute study found that interaction workers, people whose jobs involve a lot of communication and information work, spend “nearly 20 percent looking for internal information or tracking down colleagues who can help with specific tasks” [7]. Analysts, managers and engineers also lose hours checking whether numbers can be trusted.
Governance reduces that confusion and error with clear definitions, a searchable catalogue, certified data sets and visible quality scores. It also widens what people know and can do on their own, because they can find and use trusted data outside their own department.
| Objective | Typical Trigger | Governance Lever | What Success Looks Like |
| Visibility | Partners dispute orders or stock | Shared definitions and master data | One version of the truth for all parties |
| Compliance | New regulation or audit finding | Data inventory, lineage, access rules | Audit questions answered in days, not months |
| Growth and M&A | New entity, new systems | Unified definitions and reconciliation | Combined reporting without manual merges |
| Knowledge workers | Time lost validating numbers | Glossary, catalogue, quality scores | People use data instead of debating it |
The Six Dimensions of Data Quality
You cannot improve what you cannot measure. That is why most governance programs start with a shared definition of “good data”. The most widely used definition comes from the UK chapter of the Data Management Association (DAMA UK), which published The Six Primary Dimensions for Data Quality Assessment in October 2013 [8]. The UK Government Data Quality Framework (2020) uses the same six dimensions [9].
Here is what each dimension means in plain words:
- Accuracy: The data reflects reality. The customer’s address on file is the address where they actually live.
- Completeness: All required values are present. Every invoice has a tax ID, every order has a delivery date.
- Consistency: Values do not conflict within a record or across systems. The CRM and the ERP show the same customer status.
- Timeliness: The data is available when it is needed and is current enough for the decision. Yesterday’s stock level is fine for a monthly report, not for a same-day dispatch.
- Validity: The data conforms to the expected format, type and range. A date is a real date, a PIN code has six digits.
- Uniqueness: Each real-world thing is recorded once. One customer, one customer record.
The dimensions interact. Filling blanks with guesses raises completeness and lowers accuracy. A value can be perfectly valid and still wrong. Two systems can be perfectly consistent and both be inaccurate. That is why a governance team scores data on all six, not just the easiest one.
| Dimension | Typical Failure | How to Measure | Strategic Implication |
| Accuracy | Wrong address, wrong price, wrong owner | Sample checks against a trusted source | Bad decisions made with confidence |
| Completeness | Missing tax IDs, blank industry codes | % of required fields populated | Reports that silently undercount |
| Consistency | CRM and ERP disagree on revenue | Cross-system reconciliation rules | Leaders lose trust in every report |
| Timeliness | Data lands days after it is needed | Latency from event to availability | Decisions made on yesterday’s reality |
| Validity | Text in date fields, out-of-range values | Rule-based format and range checks | Broken pipelines and failed loads |
| Uniqueness | Same customer stored three times | Duplicate rate after matching | Inflated counts, wasted outreach |
Who Owns Data Quality: The Data Governance Team
The six dimensions are the scorecard. The team is what makes the score move. A typical governance team brings together four groups, each with a different job. Leadership matters most at the start: the UK Government Data Quality Framework says that “Senior leaders should be advocates for the framework in their departments” and that “Data leaders should guide an organisation or team’s strategic direction” [9]. The day-to-day work of data stewardship is covered in depth in the DAMA Data Management Body of Knowledge [10].
Executive Leadership
- Sponsorship: Gives the program authority to change processes that cross departments.
- Priorities: Decides which data domains matter most, such as customer, product or finance.
- Escalation: Settles disputes when two departments define the same thing differently.
Project Management
- Delivery: Turns the governance plan into scoped work with dates and owners.
- Coordination: Keeps IT, business teams and vendors moving in the same direction.
- Reporting: Tracks progress against the quality scores, not just tasks completed.
Line-of-Business Managers
- Business rules: Define what “correct” means for their area, such as when an order counts as shipped.
- Impact: Explain what bad data costs their team in time, money or customer trust.
- Adoption: Make sure their people follow the new standards in daily work.
Data Stewards
- Day-to-day ownership: Monitor quality scores for a data domain and fix issues at the source.
- Definitions: Maintain the business glossary so a term means the same thing everywhere.
- Bridge: Translate between business users who know the meaning and engineers who know the systems.
| Role | Pain Point Today | Governance Responsibility | Business Outcome |
| CFO / Finance head | Numbers differ across systems | Own finance data definitions | Faster, defensible close |
| COO / Operations head | Inventory and supplier data out of date | Set timeliness targets for ops data | Fewer stockouts and delays |
| Sales / Marketing head | Duplicate leads and accounts | Own customer master rules | Cleaner pipeline, less wasted outreach |
| CIO / CTO | Teams build conflicting fixes | Provide shared tools and platforms | Lower cost, one source of truth |
| Data steward | No authority to change upstream | Monitor and resolve domain issues | Problems fixed once, at the source |
The Methodology: Using Six Sigma to Improve Enterprise Data
A team needs a repeatable way to work. Many governance programs borrow Six Sigma, a quality management method first developed for manufacturing. Its DMAIC cycle, which the American Society for Quality describes as “a structured problem-solving approach used to improve existing processes” [11], maps neatly onto data:
- Define: Pick one data domain and one business problem, such as duplicate customer records hurting collections.
- Measure: Score the domain on the six dimensions to set a baseline.
- Analyze: Trace each defect to its root cause, such as a missing validation rule in an entry form.
- Improve: Fix the cause, not just the symptom. Change the form, the integration or the process.
- Control: Monitor the score continuously so the problem does not return.
Where Six Sigma Hits Its Limits
- Overhead: Full Six Sigma rigour can feel heavy for small teams. A lighter version of DMAIC is often enough.
- Fuzzy defects: Manufacturing defects are easy to count. Data defects depend on business rules that people may still be debating.
- Culture: The method fails if leaders treat it as an IT project. Business owners must own the Define and Improve steps.
Six Sigma is a common choice, not the only one. Some teams use Lean, Agile sprints or the data quality cycle described in the DAMA Data Management Body of Knowledge (DAMA-DMBOK) [10]. What matters is having a cycle and sticking to it.
The Toolset: Mapping, Profiling, Cleansing and Monitoring
Tools do not create governance, but they make it practical at scale. Most programs rely on four types of tooling, used in order.
| Step | What It Does | Typical Output | Strategic Implication |
| 1. Data Mapping | Traces where data comes from and where it goes | Source-to-target maps, lineage diagrams | You know which system to fix |
| 2. Data Profiling | Scans data to find patterns, gaps and outliers | Quality scores per dimension | You fix the worst problems first |
| 3. Data Cleansing | Corrects, standardises and de-duplicates records | Clean, matched master records | Reports become trustworthy |
| 4. Data Monitoring | Checks quality rules continuously | Alerts, dashboards, trend lines | Problems are caught before users see them |
Think of mapping as the plumbing diagram, profiling as the water test, cleansing as the filter and monitoring as the meter that tells you when the water turns bad again.
Where Tools Hit Their Limits
- No ownership, no fix: A profiling tool can flag 40,000 duplicates. It cannot decide which record is the master. A data steward must.
- Cleansing is not prevention: Cleaning data downstream without fixing the entry point means paying for the same clean-up every month.
- Tool sprawl: Buying separate tools for each department recreates the silo problem the program is meant to solve.
Why Departmental Fixes Fail, and What a Corporate Data Authority Changes
Most companies try to fix data quality one department at a time before they try governance. Those attempts usually worked locally and failed globally. Finance cleaned its customer list. Sales cleaned a different copy. The two lists still disagreed.
The fix is a corporate data authority: one person or group with the final say on data rules. When finance and sales disagree about what counts as an active customer, the authority settles it once, for the whole company. It does not replace departmental expertise. It gives that expertise a single place to land.
| Aspect | Departmental Approach | Enterprise Governance | Strategic Implication |
| Ownership | Each team owns its copy | Named stewards per domain | Clear accountability |
| Definitions | Local, often conflicting | One shared glossary | Reports reconcile |
| Fixes | Downstream, repeated | At the source, once | Lower long-run cost |
| Tools | Bought per department | Shared platform | Less sprawl, easier audit |
| Disputes | Escalate informally or not at all | Settled by the data authority | Faster decisions |
Which approach should you go for?
- Start small: Pick one high-value domain, such as customer or finance, and prove the model there.
- Name owners before buying tools: A steward with a spreadsheet beats an expensive platform nobody owns.
- Measure from day one: Baseline the six dimensions so you can show progress to leadership.
- Do not: Launch a company-wide program with no sponsor, no budget and no first domain. It will stall within a quarter.
- Do not: Treat governance as a one-time clean-up. Without monitoring, quality decays again.
What You Get from a Working Governance Program
- Reports leaders trust: Finance, sales and operations present numbers that reconcile.
- Faster audits: Lineage and access records are ready when regulators or auditors ask.
- Smoother integration: New entities and systems join a known framework instead of creating new silos.
- Less rework: Errors are fixed once, at the source, instead of in every report.
- AI-ready data: Models and copilots are only as good as the data behind them. Governed data lowers the risk of confident but wrong answers.
A Three-Phase Roadmap
Phase 1: Assessment
Map your critical data flows, profile one or two priority domains and score them on the six dimensions. Agree on the sponsor, the first stewards and the decision rights of the data authority.
Phase 2: Execution
Fix root causes in the priority domain using a DMAIC cycle. Publish the glossary, set up master data rules and build a quality dashboard that business owners actually look at.
Phase 3: Optimization
Automate monitoring and alerts, extend the model to the next domain and link quality scores to business KPIs so the program keeps its funding.
Conclusion
Data governance initiatives exist because local fixes do not scale. Six dimensions give you a shared scorecard. A cross-functional team gives the scorecard an owner. A method and a toolset keep the improvements in place.
The question is not which data quality tool to buy. It is who has the final say when the numbers disagree. Answer that, and the tools start to pay off.
So, what do data governance initiatives actually fix? They fix accountability, and better data follows.
FAQs
Sources
- Gartner. Data Quality: Why It Matters and How to Achieve It. gartner.com
- Beth Anne Cusack (RC Partners LLC). Data Governance and Data Quality: Applying GS1 Best Practices to USAID/GHSC-PSM. USAID Global Health Supply Chain Program – Procurement and Supply Management, February 2017. ghsupplychain.org
- GS1. Data Quality. GS1 Global Data Synchronisation Network. gs1.org
- Data Protection Commission (Ireland). Principles of Data Protection. dataprotection.ie
- Basel Committee on Banking Supervision. Principles for effective risk data aggregation and risk reporting (BCBS 239). Bank for International Settlements, 9 January 2013. bis.org
- Rahul Awati. How governance maturity affects M&A integration outcomes. TechTarget, 8 April 2026. techtarget.com
- Michael Chui, James Manyika, Jacques Bughin et al. The social economy: Unlocking value and productivity through social technologies. McKinsey Global Institute, July 2012. mckinsey.com
- DAMA UK Working Group. The Six Primary Dimensions for Data Quality Assessment. October 2013.
- Government Data Quality Hub. The Government Data Quality Framework. GOV.UK, 3 December 2020. gov.uk
- DAMA International. DAMA-DMBOK: Data Management Body of Knowledge, 2nd edition. Technics Publications. ISBN 978-1-63462-234-9.
- American Society for Quality. DMAIC Process: Define, Measure, Analyze, Improve, Control. asq.org
Author By
Gaurav Lakhotia
Gaurav holds an MBA in Business Analytics. As a data geek and avid learner, he has earned accolades including Microsoft Certified Data Analyst, Microsoft Certified Azure Administrator, and active Power BI Community contributor. He is also a Microsoft Certified Trainer. With critical thinking and project delivery skills, he strives to achieve customer delight in Data Analytics projects. Gaurav loves travelling with friends when he is not busy working on data.
