Utility Sub-Metering
Auditors Stopped Hunting and Started Deciding
The Outcome
The team works from a prioritized queue instead of raw spreadsheets. Every read gets screened, reviewers spend their time on the ones that need a human decision, and field crews get consistent work orders out of the same system.
Staff read every CSV row by hand
Every read screened automatically
Results depended on who reviewed
One standard, and every call logged
Work orders written up separately
One click from an approved finding
The Bottleneck
Meter reads came in as big CSV exports and staff went through them by hand. Actual problems were buried in thousands of normal reads, results depended on who was doing the review, and writing up work orders for the field was another manual step on top of that.
What We Built
- Screening on every read as it lands, using business rules plus a machine learning model, so nothing waits on someone getting to it
- A reason attached to every flag, so reviewers can trust the queue instead of re-checking it
- A review queue that keeps a human in the loop: approve, override, or escalate before anything reaches the field
- Work orders generated from approved findings, with assignment and status tracking already attached
- Multi-tenant separation, so each property portfolio can be handled without cross-contamination
Fixed-scope platform buildOngoing support & optimizationClient owns the infrastructure