Automating Public Utility Meter Anomaly Detection in North America

Background

Our client is a municipally owned public utility in Florida established in 1923. It serves as one of the largest municipal utilities in Florida, providing electric, water, and more to over 240,000 customers. They operate a diverse generation portfolio and are actively pursuing clean energy initiatives.

Challenge

Utility companies deal with vast amounts of metering data on a daily basis, making it difficult to manually identify irregularities across their customer base. Our client faced the challenge of detecting anomalous consumption patterns - specifically meters showing zero consumption, as well as those with unusually low or high usage compared to expected norms. Left undetected, these anomalies can signal anything from faulty meters and data quality issues to potential theft or undetected leaks, all of which have real operational and financial implications.

Approach

To tackle this challenge, we leveraged SQL logic within Snowflake to identify and flag anomalous consumption patters across the meter data. We built out views directly in Snowflake as well as utilizing Talend for data integration and transformation, ensuring the data was clean and structured before any analysis took place. The anomaly detection focused on three key categories: - Zero Consumption: identifying meters recording no usage over a defined period - Low Usage: flagging meters with consumption significantly below expected norms - High Usage: detecting meters with unusually elevated consumption To bring the findings to life, we developed an exploratory Tableau dashboard that allowed us to visually surface these anomalies - making it easy to spot patterns that would have been impossible to catch manually in the raw data.

Key results

Benefits

'The work delivered real, tangible value to our client by transforming a manual, time-consuming process into an automated, data-driven solution. By proactively identifying zero consumption and abnormal usage patterns, the client was able to reduce revenue loss, minimize operational risk, and ensure their metering data could be trusted for downstream decision making. Beyond immediate findings, the Snowflake views, SQL logic, and Tableau dashboard created a reusable foundation that the client can build on - whether that's expanding anomaly detection to new use cases or laying the groundwork for more advanced predictive analytics down the line. Ultimately, this project demonstrated that with the right data infrastructure and exploratory mindset, utilities can turn raw metering data into a powerful operational tool.

Share this key play