34% NRW reduction in six months — WATCO Odisha.

Water lost. Location unknown. NRW above 50%.

Non-revenue water was WATCO Odisha's single biggest operational cost — at over 50% of total production, more than half of every litre treated was being lost before it reached a paying consumer. The utility knew water was being lost. It did not know where, how much per zone, or who was responsible.

Physical leak surveys, manual meter reading, and periodic spot checks had been running for years without meaningful improvement. The data needed to identify where losses were occurring did not exist in a structured form that could drive prioritised field action.

• NRW above 50% — half of every litre treated lost before paying consumer
• Loss location unknown — could not prioritise field investigation
• Manual meter reading — billing accuracy and disputes worsening
• No DMA-level water balance — losses invisible at zone granularity

A knowledge graph of every pipe segment.

Cognecto deployed AMR smart metering, segmented the 120 km network into District Metered Areas, and built a Cortex knowledge graph linking every sensor reading, every consumption anomaly, every billing discrepancy to the pipe segment and district it came from.

NRW was no longer a single percentage — it was a map of loss severity per zone, ranked by intervention value. Minimum Night Flow analysis isolated background leakage. Field crews finally had data to direct investigation.

"NRW fell 34% in the first six months. Billing accuracy improved 28%."

Both outcomes from the same data, structured correctly for the first time. The water network at WATCO Odisha did not change. The infrastructure did not change. What changed was the data layer underneath — and the operational decisions that data made possible.

The numbers behind the story.