Data Drop - Clean
Automatically checks & repairs 98% of your data
Turns scattered sources — SCADA, billing, measurements, break history — into one consistent, validated data basis your team can actually trust and act on. It automatically cleans, harmonizes, and governs the data — identifiers, units, time bases — to produce a continuously updated single source of truth, built in an easy-to-access data lakehouse architecture.
Book a meetingHow it works
A clear four-layer architecture
-
Ingestion
Pulls in SCADA, GIS, billing and break history, plus uploads for offline data. Connectors handle each source automatically; manual uploads cover measurement campaigns or offline sensors utilities may still run alongside their core systems. Everything ingested is data the utility already collects — nothing new to install, nothing new to maintain.
-
Storage
A data lakehouse stores raw and curated datasets, preserving full traceability. Data is stored economically as Parquet files, with the original raw data always kept intact — so nothing is ever silently overwritten. Alongside it, curated, analytics-ready tables are exposed for downstream tasks like leak detection or hydraulic modeling, without touching the source of truth underneath.
-
Governance
Validates and repairs every incoming record automatically, harmonizing identifiers, units and time bases so all sources speak the same language. Every correction is logged, keeping the data auditable and the single source of truth continuously up to date.
-
Serving
Dashboards and APIs expose the platform, built to fit existing workflows. Early deployments can run on Grafana, evolving into a dedicated web application as a utility's needs grow — the point isn't to replace how your team already works, it's to plug a trustworthy data layer into it.
“Most leak-detection and prevention approaches fail in practice not because the algorithms are weak, but because the input data is fragmented and unreliable. Data Drop makes the data trustworthy and auditable, so advanced leak analytics becomes operationally usable at scale.”