Define who decides about data and how a working integration will be accepted

Data Governance and Integration: Connect Ownership with Data Flows

Conflicting metrics and unclear data origins need more than a new interface.

  • 01Assign decisions about data use, meaning and ownership
  • 02Describe sources, transformations and targets with testable rules
  • 03Accept a pilot against agreed quality, exception and change controls
11+Years of experience
120+Employees
540+Projects
ISO 27001certified

What governance and integration each need to resolve

Data governance establishes who may decide about data use, definitions, quality requirements and exceptions. Data integration implements the agreed flow: which source supplies each item, how it changes and where it is used. Connecting these tasks helps a team explain a result and resolve an incorrect or incomplete handover.

Start with a business question and the actual source and target systems. Agree definitions, ownership, permitted use and suitable checks before choosing an integration pattern or catalog tool. Technical metadata can help trace relationships, but business meaning and manual processing also need confirmation. A platform does not replace those decisions. The monthly-revenue example below shows how a business rule becomes a data-flow requirement and an acceptance case.

An engagement can cover an initial assessment, governance and data-flow design, selected implementation work and an agreed pilot handover. Define the data domain, sources, targets, required controls and expected outputs in the work package. Your organization provides business owners, permitted representative data and the relevant system contacts. Access approvals, software procurement, ongoing operation and expansion to other domains require an explicit agreement.

Define the result before selecting the tool: Agree what the data must mean and how the result will be checked. Catalogs, integration platforms and quality tools can support the work; their presence alone does not establish correct reporting or regulatory compliance.

3 service modules

Define data ownership, implement selected flows and verify quality

01

Governance and business definitions

We define the decisions, roles and rules needed for the selected data use. The output connects a business definition to its owner and to the metadata required for implementation.

  • Agree the use case, data meaning and relevant domain boundaries
  • Assign ownership, stewardship, approval and exception responsibilities
  • Record definitions, sources and permitted-use decisions
02

Integration and traceable data flows

The selected flow is designed or implemented against agreed source and target behavior. Transformation, timing and error handling remain visible so the result can be explained and maintained.

  • Map source fields, transformations, keys and target outputs
  • Select an integration pattern against volume, latency and system constraints
  • Describe lineage, replay or correction handling and change dependencies
03

Quality checks and operational handover

A pilot checks the business result and the agreed handling of incomplete or incorrect data. Findings distinguish demonstrated behavior from assumptions or unresolved issues.

  • Define expected outcomes and representative exception cases
  • Assign issue handling, quality review and change approval
  • Record acceptance evidence and responsibilities after handover

Example: Three rules for a monthly revenue report

This original illustrative worksheet connects business use, data integration and ownership. Data, organization and decisions are fictional; actual thresholds and approvals must be agreed for the real report.

Completed data-flow and acceptance worksheet. On small screens, scroll the table horizontally to read all columns.
Business useRule and data flowAccountable decisionTest caseCompletion check
Show revenue for the confirmed reporting monthUse the approved ERP definition of the reporting period; agree the cutoff and time zoneFinance confirms the period definition and treatment of later postingsCheck postings before, at and after the agreed cutoffAssignment and documented exceptions match the approved rules
Combine revenue in different currenciesKeep amount and currency linked; use the agreed rate source and conversion timeThe business owner decides the reporting currency, rate and rounding rulesCheck a case per currency plus a missing rate and the rounding boundaryTotals are explainable; a missing rate produces the agreed exception handling
Assign revenue to a customer groupMap source keys to the target using the approved relationship; make multiple matches visibleThe data owner confirms customer mapping and treatment of unknown keysTest a unique, an unknown and a multiply matched keyThe report has no unexplained multiplication; unresolved mapping has an accountable decision

5 phases

From one business question to an accepted data flow

A bounded pilot makes definitions, implementation choices and quality checks reviewable before broader rollout.

  1. Agree the task

    Identify the business result, data domain, sources and decision owners.

    a defined review and delivery boundary.

  2. Define meaning and ownership

    Confirm terminology, quality expectations and authorized data use.

    approved definitions and responsibilities.

  3. Design or implement the flow

    Document mappings, transformations, timing and exception handling.

    a traceable flow for the agreed systems.

  4. Check the pilot

    Compare representative results and exceptions with pre-agreed expectations.

    test evidence and a decision on unresolved issues.

  5. Hand over

    Confirm acceptance, issue ownership and change controls.

    an agreed operating handover and any remaining limits.

Asan Stefanski

Your contact

Asan Stefanski

Head of Digital Transformation

11+ years of experience, Applied Computer Science degree, Strategic planning and management of AI projects, Cyber Security, Secure Software Development, AI

Effective data governance is not a one-time initiative, but a continuous process. The keys to success are clear processes, defined responsibilities, and the right balance between control and flexibility.
ADVISORI expert opinion

11 QUESTIONS, BRIEFLY ANSWERED

Frequently asked questions about Data Governance & Data Integration

What is data governance?

Data governance establishes decision rights, accountability and rules for working with data. Within a particular domain, it should be clear who confirms definitions, permitted use, quality and exceptions. Its value depends on those decisions reaching the actual data process. A list of roles or a catalog alone changes neither the data nor the handling of a problem.

How do you develop a suitable data-governance framework?

Start with a bounded business purpose and the data it needs. Review current problems, rules and participating roles. Then agree decision rights, terminology, quality requirements and exception paths. Exercise the model through an actual data flow and adjust it against the findings before extending it to further domains.

How does data governance differ from data management?

Governance determines which decisions are needed and who may make them. Data management implements the agreed rules in collection, storage, processing, quality and use. The work is connected: a technical change may need business approval, while a governance rule must be workable in the affected systems and processes.

How can heterogeneous data sources be integrated?

First clarify meaning, use and required timeliness. Describe mappings, transformations, interfaces, access and exception handling. Choose ETL, ELT, API-based or event-based processing against the task and environment; a pattern name alone guarantees no particular freshness. Check the business result as well as technical execution, especially where keys, periods or units differ.

Which roles are needed?

Key responsibilities include business ownership of definitions and use, day-to-day handling of data questions, and technical ownership of systems and flows. Titles such as Data Owner or Data Steward help only when authority and accessible contacts are clear. One person may hold several responsibilities; decisions, implementation and review must still be traceable.

How is success measured?

Derive measures from the selected business purpose. Examples include an explained reporting difference, reliably resolved data errors or demonstrated timeliness. Record the population, observation time and starting position. Counts of catalog entries or appointed roles show activity; they do not automatically demonstrate better decisions or fulfillment of a legal requirement.

What determines cost and duration?

Relevant factors include domains, participating systems, existing definitions, data problems, integration work and available decision-makers. A business design is a different scope from technical implementation or continuing data remediation. Agree outputs, required participation and acceptance criteria. Consider licensing and operating costs for the chosen model separately; the example worksheet does not establish a fixed delivery period.

What information is needed before an engagement?

Describe the business purpose, affected users, source and target systems and known problems. Identify people able to confirm definitions and the expected result. Descriptions and suitable examples without confidential original data are enough for an initial assessment. Any required access and data use must then be explicitly agreed.

What should acceptance of a data flow check?

Check the agreed business result, technical processing and exception handling. Relevant cases may include cutoffs, units, key mapping, completeness and repeated execution. Define expected outcomes in advance and record deviations. The worked matrix connects such cases with an accountable decision and a checkable completion condition.

What do data lineage and impact analysis contribute?

Data lineage records where data comes from and the processing steps leading to an output. Impact analysis uses those relationships to identify, for example, reports affected by a proposed change. Automated discovery can capture technical relationships; business meaning, unrecorded handovers and manual processing still need review. A catalog entry alone does not establish that a report is correct or that its lineage is complete.

Does a governance tool satisfy privacy or banking requirements?

A tool can support ownership, metadata and controls. The lawfulness of a particular use and applicable supervisory requirements must be assessed for the organization and task. Banks and other regulated organizations need suitable evidence for their applicable obligations. A catalog or data-flow diagram alone does not establish full regulatory compliance.