# Building a data governancedata governanceData governance is the set of policies, roles, and processes that ensure data is accurate, secure, well-defined, and used responsibly across an organization.View full definition → council for a utility
A transformertransformerA Transformer is a neural network architecture that uses self-attention to process sequences in parallel, powering most modern language and generative AI models.View full definition → fails at 2 a.m. The outage management system pulls asset age from one database, maintenance history from another, and customer count from a third. The three numbers do not agree. Metering says the transformertransformerA Transformer is a neural network architecture that uses self-attention to process sequences in parallel, powering most modern language and generative AI models.View full definition → serial number was retired in 2019. Grid operations says it is still energized and carrying load. Nobody owns the disagreement. This is not a hypothetical: mismatched asset records are one of the most common root causes of delayed outage restoration audits at US investor owned utilities, according to postmortems regularly cited in NERC (North American Electric Reliability Corporation) reliability reports. The fix is not a better database. It is a governance body with the authority to force a decision.
Utilities sit on data with three competing owners by design:
These three views describe the same real-world object but live in different systems: a Meter Data Management System (MDMS), a Customer Information System (CIS), and a Geographic Information System (GIS) or Asset Management System (often built on tools like IBM Maximo or SAP PM). When a field crew replaces a transformertransformerA Transformer is a neural network architecture that uses self-attention to process sequences in parallel, powering most modern language and generative AI models.View full definition →, someone has to update all three, in the right order, or downstream systems (billing, outage maps, load forecasting) silently drift apart.
No single team has authority to force the other two to fix their side. That is the governance gap.
A data governance council (sometimes called a data stewardshipdata stewardshipA business-side owner responsible for the quality, consistency and appropriate use of data in their domain.View full definition → board) is a standing cross-functional body with formal authority to:
1. Define who is the system of record for each data domain (the one authoritative source when copies disagree).
2. Approve or reject changes to shared data models (for example, adding a new "asset status" field that both GIS and CIS must respect).
3. Adjudicate disputes when two teams both claim ownership or both claim veto power.
4. Set data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.View full definition → thresholds and sign off on remediation plans.
This is distinct from IT governance (which manages systems and infrastructure) and from a compliance function (which manages regulatory obligations). The council owns the *business meaning and authority* over data, not the pipes it flows through.
A charter that skips any of these four elements will fail within a year.
1. Scope. Name the data domains explicitly: asset master data, meter point data, customer data, outage data. Vague scope ("all utility data") produces a council that never meets because nobody knows if their issue qualifies.
2. Decision rights. Use a simple RACI-style split (Responsible, Accountable, Consulted, Informed) per domain. Example:
| Domain | System of record | Accountable owner | Consulted |
|---|---|---|---|
| Physical asset attributes | GIS/Asset Management | Grid Operations | Metering, Engineering |
| Meter point status | MDMS | Metering | Commercial |
| Customer-to-meter linkage | CIS | Commercial | Metering, Billing |
3. Escalation path. Define what happens when the two accountable owners disagree, for instance grid operations says a transformertransformerA Transformer is a neural network architecture that uses self-attention to process sequences in parallel, powering most modern language and generative AI models.View full definition → is decommissioned but metering still shows active reads. The path typically runs: data stewarddata stewardA business-side owner responsible for the quality, consistency and appropriate use of data in their domain.View full definition → level (weekly working group) to council level (monthly) to executive sponsor (CIO, CDO, or COO) if unresolved after a set number of days, commonly 10 to 15 business days in mature programs.
4. Cadence and quorum. Monthly council meetings, weekly steward triage, defined quorum (for example, at least one accountable owner from each domain plus a compliance or privacy representative).
Governance decisions in the US and EU do not happen in a vacuum; they sit inside binding rules.
The council's escalation path should always route privacy-flagged disputes (does this dataset contain personal data, can it be shared across teams) to the compliance representative before a business decision is finalized. For a plain-language primer on the GDPR provisions most relevant to utilities, see the European Data Protection Board's guidance.
Back to the 2 a.m. outage. Here is how the council process resolves it in practice:
1. Field technician flags the mismatch via a data qualitydata qualityThe degree to which data is fit for purpose: accurate, complete, consistent, timely, valid and unique. Poor quality data undermines analytics, reporting and AI.View full definition → ticket.
2. Data stewards from metering and grid ops triage weekly: grid ops confirms via GIS that the transformertransformerA Transformer is a neural network architecture that uses self-attention to process sequences in parallel, powering most modern language and generative AI models.View full definition → was physically replaced in 2019; metering confirms its record still points to the old serial number.
3. Root cause: the decommissioning workflow never triggered an automated update to MDMS. This is a process gap, not a data entry error.
4. Council decision (per the RACI table): GIS is system of record for asset status, so MDMS must be corrected to match. But the council also mandates a system integration fix: decommissioning events in GIS must publish an automated event to MDMS going forward.
5. Executive sponsor approves budget for the integration fix within the next sprint cycle.
A simple check any analyst can run before escalating a dispute like this is a record reconciliation query:
SELECT g.asset_id, g.status AS grid_status, m.asset_id, m.status AS meter_status
FROM gis_assets g
FULL OUTER JOIN mdms_assets m ON g.serial_number = m.serial_number
WHERE g.status != m.status OR g.status IS NULL OR m.status IS NULL;This kind of reconciliation check, run on a schedule (daily or weekly), is the practical audit that feeds the council's agenda rather than replaces it.
Knowledge check
1. In the transformer outage scenario, why does mismatched asset data across systems persist unresolved even after the discrepancy is discovered?
2. Why do utilities have a particularly acute need for formal data governance compared to organizations with simpler data structures?
3. A field crew replaces a transformer. Two weeks later, the GIS shows the new asset, but the outage management system still references the old serial number. What does this scenario best illustrate?
4. Select ALL correct answers about the three competing data owners at a utility described in the lesson.
Select all the correct answers.
5. Select ALL correct answers about what a data governance council is designed to do.
Select all the correct answers.
What is Data Governance?