Data Quality Management: Tasks, Process and Roles

What Is Data Quality Management (DQM)?

Data quality management – DQM for short – covers all the rules, roles, processes and tools a company uses to secure the quality of its data permanently. It answers three questions: What quality do our processes need? How good is the data actually? And who makes sure it stays that way?

The difference from data quality is the difference between a state and a task. Data quality describes how complete, accurate and consistent data is – our guide to improving data quality shows which criteria matter and which measures improve that state. Data quality management is the ongoing task of measuring, steering and maintaining that state. For manufacturers, a familiar comparison fits: what quality management does for products, data quality management does for data.

What Are the Tasks of Data Quality Management?

At its core, there are six tasks:

• Define requirements: Which data is critical for which processes, and what quality does it need?
• Define rules: Requirements turn into testable data quality rules, such as "Every business customer has a valid VAT ID".
• Measure and monitor: Metrics show how many records meet the rules – once at the start and continuously afterwards.
• Fix errors and their causes: Wrong records are corrected. More important is eliminating the cause so the same error does not arise again.
• Assign responsibility: For every data type, it is clear who sets the rules and who takes care of maintenance day to day.
• Report: Management and business departments regularly see where data quality stands and whether it is improving.

The Process: Data Quality Management as a Cycle

Many process models go back to the Total Data Quality Management (TDQM) approach at MIT. In 1998, Richard Y. Wang described it as a cycle of four phases, modeled on the Deming cycle from quality management [1]:

Then the cycle starts again. New systems, new processes and simply time change the data: customers move, items are discontinued, contacts change. Other models divide the phases more finely, for example with separate steps for monitoring and reporting – the core remains the same. Data quality management is therefore not a project with an end date but a control cycle.

Data Quality Rules: Making Quality Measurable

A data quality rule is a statement that can be checked automatically with yes or no for every record. Good rules are justified by the business and technically testable:

Each rule becomes a metric: the share of records that meet it. With a target value and a warning threshold, this creates a traffic light that shows where action is needed. The selection is decisive, because not every field is equally important. Start with the critical data elements – the fields without which an invoice, a delivery or a payment fails. Our guide to improving data quality describes how to collect the metrics in its section on measurement.

Roles: Who Is Responsible for Data Quality?

In our experience, data quality management rarely fails because of technology, but because of unclear responsibility. A simple distribution of roles has proven itself:

• Management: sets the goals, releases time and budget and reviews the metrics regularly.
• Data owner: is responsible for a data type from the business side – for example, the head of sales for customer data – and decides on rules and target values.
• Data steward: takes care of maintenance and checks day to day, resolves error cases and trains colleagues.
• Everyone who enters data: makes sure data gets into the system correctly the first time.
• IT: implements validation rules, interfaces and reports technically – but is not responsible for the content.

A common mistake is to hand data quality to IT alone. IT can enforce a rule technically, but only the business department knows whether an address or a price is correct. In mid-sized companies, the roles are therefore not new positions but clearly assigned tasks of existing employees. A dedicated data quality manager position usually only pays off with very large datasets. Our data governance guide shows how the role model fits into the rest of data governance.

Distinguishing Data Quality Management, Data Governance and Data Management

The terms are often mixed up. This is how they relate:

Data quality management thus implements the quality requirements of data governance: it makes quality measurable and keeps it there. It is most effective with master data – more on this in our article on master data management. Our article on the data management system describes the framework for data management as a whole.

Which Software Supports Data Quality Management?

For each phase of the cycle, there are suitable software groups:

• Define: A data catalog documents terms, rules and owners in one place.
• Measure: Data profiling and monitoring tools check the rules automatically and alert you when a metric drops below the threshold. Newer tools group this under the term data observability.
• Analyze: Data lineage shows which systems and interfaces an incorrect value has passed through. Our smart data catalog combines both: automatic metadata capture and data lineage.
• Improve: Duplicate detection and matching, validation rules in CRM and ERP and AI-assisted enrichment fix errors and prevent new ones.

To get started, less is often needed than expected: one query per rule, an overview of the most important metrics and a fixed date on which error cases are resolved.

Introducing Data Quality Management: in Five Steps

Data quality management does not have to start as a big program. A step-by-step approach has proven itself:

1. Choose one data type: the one that causes the most damage – usually customer or item master data.
2. Define critical fields and rules: five to ten rules are enough to start with.
3. Name the people in charge: a data owner and a data steward for this data type.
4. Measure and make visible: collect the metrics once and then report them monthly as a traffic light.
5. Eliminate causes and expand: introduce validation rules at the source, then tackle the next data type.

Our case study from a machinery manufacturer shows what this looks like in practice: a data quality audit, a central master data registry, automated checks in the ERP workflows and team training – exactly this cycle.

Where does your data quality management stand? With our data maturity assessment, we measure quality, silos and ownership and deliver a prioritized roadmap. The data hygiene audit shows where your data needs cleaning. The fastest way is a direct conversation: book a 30-minute intro call.

Data Quality Management and AI

AI applications need not just clean data once, but data that stays clean. An assistant that gives correct answers today gives wrong ones as soon as the data beneath it becomes outdated. For high-risk AI systems, the EU AI Act even prescribes data quality: Article 10 requires data governance practices as well as training, validation and testing data that are relevant, sufficiently representative and, to the best extent possible, free of errors and complete [2]. Working data quality management lays the foundation for this – our EU AI Act readiness clarifies where you stand on the other obligations.

Frequently Asked Questions About Data Quality Management

What Is the Difference Between Data Quality and Data Quality Management?

Data quality is the state of the data – how complete, accurate and consistent it is. Data quality management is the ongoing task of measuring, improving and maintaining that state.

What Is the Difference Between Data Management and Data Quality Management?

Data management is the umbrella term for everything to do with capturing, storing, connecting and using data. Data quality management is the part of it that takes care of quality: defining rules, measuring, fixing errors and clarifying responsibility.

Which Processes Does Data Quality Management Include?

The cycle of define, measure, analyze and improve, plus continuous monitoring and regular reporting. The standard ISO 8000-61 structures the processes similarly: data quality planning, control, assurance and improvement, likewise as a cycle following the PDCA principle [3].

How Is Data Quality Management Implemented?

Step by step: choose one data type, define rules, name the people in charge, measure and eliminate causes – then the next data type. The five steps above are a proven way to start.

Does a Mid-Sized Company Need a Data Quality Manager?

Usually not as a dedicated position. It is more important that a data owner and a data steward are named for every important data type – and that they get time for it. If that time is not available internally, the role can also be filled externally for a while.

Are There Standards for Data Quality?

Yes. ISO 8000 is the international series of standards on data quality: part 61 describes a process reference model for data quality management, other parts govern the quality of master data exchanged between organizations [3]. ISO/IEC 25012 defines a data quality model with 15 characteristics, divided into inherent and system-dependent quality [4]. For mid-sized companies, both serve as orientation rather than obligation.

Conclusion

Data quality management sounds like big business, but at its core it is simple: define the quality the most important data needs, measure regularly, eliminate causes and assign responsibility clearly. Anyone who starts with one data type and a few rules can have a first working control cycle within a few weeks – and with it the foundation for reliable processes, reports and AI.

References

1. (Wang, R. Y. – A Product Perspective on Total Data Quality Management, Communications of the ACM 41(2), 1998)
2. (Regulation (EU) 2024/1689 (AI Act), Article 10 – Data and data governance)
3. (ISO 8000-61 – Data quality management: Process reference model)
4. (ISO/IEC 25012 – Data Quality Model)