LET'S CONNECT

Your Data is Lying to You: Clean Data Strategies for Utilities

January 2026
  • In previous blog posts, we’ve talked about the importance of data in leveraging AI responsibly, how meta data drives trustworthy AI, and the part that data plays in managing AI risk. But as we seek to realize the promise of AI - or any of the other modern technologies that promise to revolutionize the electric utility - we need to face a fact we all know; for most utilities, the data isn’t going to get us there.

  • There are a lot of good reasons that utilities struggle with data. In a business that operates assets that are expected to last up to 50 years, we should expect that the criticality of accurate data might be more important today than it was in, say, 1975. Case in point:
  • We’ve seen examples of entire neighborhoods that are on the feeder’s B phase according to GIS data and the network model used by the ADMS/Outage system
  • We’ve seen fleets of meters that all have the exact same physical location, but serve an entire avenue of customers
  • We’ve also seen meters that are documented in the system as being located in the middle of the ocean.
  • In the time of bell bottoms and disco, when some of this data was created, nobody would really have noticed other than the most persnickety of foreman and distribution engineers.
Obviously, in 2025 this all looks a lot different. If our data were our child’s bedroom it would be in complete disarray. It would resemble the bedroom of a teenager, with dirty clothing strewn all about, and dishes that have been missing from the kitchen for weeks.
If we think that our data will underpin AI strategies, get ready for some really wild results. The lack of data quality makes it impossible to distinguish hallucinations from earnest data-based responses. The good news is that other industries have addressed similar problems, and proven maturity models exist to help us improve our data and realize the benefits of AI and data-driven insights without waiting for weeks or investing in transformation projects to help us get there.
A Maturity Model Provides a Starting Point for Data Improvement
If we are going to improve our data, we need to know what "good" looks like. A maturity model can help us figure out where we are right now, and what benefit we would get from getting to the next step. In other words, we need to identify the “why.” We provide a data maturity framework the helps a utility identify their level of maturity, specific steps to progress to the next level, and metrics utilities can realistically track and utilize with existing systems and processes.

The Utility Data Maturity Model

The Utility Data Maturity Model characterizes the current state of a utility’s data against defined criteria and provides a framework for incremental (and sustainable) improvement.

Level 1 Assumed & Reactive

What it Looks Like: Data exists and is implicitly trusted, despite being fragmented. Issues emerge when billing, outage, or build work identify them (often reported by customers or field personnel who are impacted by the issue). Data errors causing or contributing to the problem are usually identified and fixed, but there is no mechanism to keep them in synch or prevent the problem from reoccurring.
How to Progress to the Next Level: Create shared clarity around which data matters most and who is responsible for it.
Specific Actions You Can Take
  • Identify Critical Data Elements. Focus on the data that directly supports safety, reliability, billing accuracy, and system planning. Common priorities include feeder connectivity, phase, meter location, transformer relationships, and essential asset attributes.
  • Establish Clear Ownership. Assign named owners for critical data elements who are responsible for definitions, quality expectations, and issue resolution across systems.
  • Define Data Expectations in Plain Language. Document what completeness, accuracy, and validity mean for each priority data element so teams share a common understanding.
  • Create a Consistent Way to Capture Data Issues. Track problems as they surface during operations, planning, and customer interactions to reveal recurring patterns and systemic gaps.
Example A utility begins logging feeder phase discrepancies identified during outage restoration. Over time, this creates visibility into where assumptions persist and where targeted remediation will deliver the most value.
How to Measure Readiness for the Next Level
  • Critical data elements with named owners (%) Measures whether responsibility for key data has been explicitly assigned
  • Documented data definitions coverage (%) Percentage of priority data elements with agreed-upon definitions and quality expectations
  • Data Issues Captured per Month (count) An increase initially indicates improved visibility, rather than worsening quality
  • Repeat Data Issues (%) Tracks how often the same issue appears in the same location or system

Level 2 Defined

What it Looks Like: Each system-of-record has data stewards, ownership, is part of a data dictionary, and is governed. This approach clarifies data quality expectations. It aligns with the language used by operations and engineering. Meta-data problems can be identified between systems, and automation can be used to identify problems. Automated cleanup can be done with some automation.
How to Progress to the Next Level: Focus on reinforcing data quality through everyday workflows.
Specific Actions You Can Take
  • Apply validation at the Point of Data Creation and Change. Require key attributes and relationships before work can be completed, ensuring accuracy is addressed as part of normal operations.
  • Schedule Regular Cross-system Reconciliation. Compare GIS, AMI, CIS, and planning data to identify inconsistencies early and keep systems aligned.
  • Use Automation to Surface Issues. Implement rules and analytics that flag anomalies and gaps, enabling teams to focus on resolution rather than discovery.
  • Align Operational Processes with Data Standards. Ensure field work, design updates, and system changes naturally reinforce established data expectations
Example Validating service design data before construction begins resolves connectivity issues once, improving both planning accuracy and field execution.
How to Measure Readiness for the Next Level
  • Validation Coverage at Data Creation (%) Percentage of critical data updates subject to automated validation rules
  • Cross-system Consistency Rate (%) Degree of alignment between GIS, AMI, CIS, and planning systems for key attributes
  • Automated Data Quality Checks (Count) Number of active rules detecting anomalies and gaps
  • Manual Data Correction Effort (Hours per Month) A declining trend indicates that issues are being addressed earlier

Level 3 Integrated

What it Looks Like: Data quality is embedded into workflows and systems, where it is enforced. Cross-system consistency can be measured and improved. Data resonates across the organization, including operations, engineering, and IT.
How to Progress to the Next Level: Demonstrate reliability through measurement, feedback, and transparency.
Specific Actions You Can Take
  • Track data Quality Metrics Over Time. Monitor completeness, consistency, and accuracy for priority data elements to understand trends and improvement areas.
  • Create Feedback Loops Between Analytics and Source Systems. Use insights from forecasting, planning, and operations to drive targeted data corrections where they will have the greatest impact.
  • Incorporate Data Confidence into Decision Processes. Make the level of data reliability visible so decisions reflect both opportunity and risk.
  • Share Data Quality Insights Broadly. Provide engineers, operators, and analysts with visibility into where data is strong and where assumptions remain.
Example By prioritizing feeders with low connectivity confidence for remediation during routine maintenance, reliability of planning models steadily improves without disrupting operations.
How to Measure Readiness for the Next Level
  • Data Completeness for Priority Attributes (%) Tracks required fields for assets, connectivity, and services
  • Data Accuracy Confidence Score Composite score based on reconciliation, validation, and field verification
  • Analytics Rework Rate (%) Frequency of model reruns due to data quality issues
  • Forecast Variance Attributable to Data (%) Identifies how much error stems from data versus assumptions or scenarios
  • Decisions Supported by Governed Data (%) Measures how often planning and operational decisions rely on validated sources

Level 4 Trusted

What it Looks Like: Data is reliable enough to enable automation, analytical insights, and AI/ML. The focus is on outcomes, not tooling. Data supports data-based decision making. Decisions are incorporated into day-to-day operations. Buzz words fall away from cross-functional efforts and are replaced by business-focused aspirational outcomes that can be measured.
How to Sustain & Evolve: Maintain data quality through continuous learning & adaptation.
Specific Actions You Can Take
  • Update standards as grid behavior and technologies change
  • Use analytics to identify emerging data gaps
  • Reinforce feedback between operational outcomes and data models
  • Treat data quality as an ongoing operational discipline
Example As distributed energy resources increase, a utility evolves its connectivity standards and validation rules to support more complex power flows and planning scenarios.
How to Measure Sustainment and Evolving Maturity
  • Automated Anomaly Detection Coverage (%) Portion of data monitored continuously for unexpected pattern
  • Upstream Correction Rate (%) Percentage of data issues resolved in source systems rather than downstream models
  • Time from Insight to Correction (Days) Measures how quickly analytics drive meaningful data improvements.
  • AI or Advanced Analytics Confidence Indicators Stability and explainability of model outputs over time
  • Regulatory or Stakeholder Data Challenge Rate Frequency of data-related disputes or rework in filings and reporting

Progress over Perfection

Clean data is the result of deliberate, sustained progress. Most utilities already know their data is imperfect. What often stalls momentum is the belief that meaningful improvement requires massive programs, system replacements, or years of cleanup before value can be realized. The maturity model above offers a different path. One that focuses on clarity, ownership, integration, and trust, step by step. The most successful utilities do not try to leap to advanced analytics or AI all at once. They focus on moving from one level to the next, choosing actions that reduce friction, improve confidence, and support better decisions today. Each step compounds. Each improvement makes the next one easier. The question is whether your data is improving in ways that matter. We help utilities understand where they are on this path, identify the few changes that will deliver the most value, and make progress that lasts. Not by boiling the ocean, but by aligning data, systems, and operations around outcomes that can be measured and sustained. Because when your data starts telling the truth, the rest of the work gets a lot easier.