Solar & Wind Energy
Data latency achieved, down from a once-daily CSV pull
Client NPS improvement
About
The operator set out to give every power-purchase client a near-real-time performance window into their solar and wind assets. IoT meters wrote CSVs once a day, and monthly PDF reports obscured inverter trips, sparking SLA disputes and eroding client NPS. We streamed live data through Azure IoT Hub into a token-secured portal with automated under-performance alerts.
Industry
Solar & Wind Energy
Company size
200 – 500 employees
Founded
2012
The Company
A renewable-energy IPP reporting on a monthly delay
The operator develops and operates solar and wind generation assets under long-term power-purchase agreements, where clients expect visibility into how their contracted capacity is actually performing.
IoT meters across the site fleet were writing performance data to CSVs once a day, and clients received a consolidated performance summary as a monthly PDF — a reporting cadence built for compliance record-keeping rather than the operational visibility clients actually wanted.
The challenge
A monthly PDF that hid the problem until it was old news
An inverter trip or an under-performing string could sit unaddressed for weeks before it showed up in the next monthly PDF, by which point the client had often already noticed the shortfall on their own utility bill and raised an SLA dispute.
The lag between an issue occurring and a client learning about it — and the operator addressing it — was eroding trust and client NPS scores, even in cases where the underlying issue was resolved reasonably quickly once someone actually caught it.
The Solution
Streaming telemetry into a client-facing portal with automated alerts
We streamed site telemetry through Azure IoT Hub into Delta Lake, replacing the once-daily CSV batch with continuous ingestion. A token-authenticated Tableau portal gives each client a 15-minute-refresh view of their own asset's KPIs, and an ARIMA-based under-performance detector automatically flags anomalies rather than waiting for a human to spot them in a monthly summary.
Clients now see an inverter trip or an under-performing string within roughly the same window the operator's own operations team does, closing the information gap that had been driving SLA disputes.
The Results
Latency down to 10 minutes, NPS up 12 points
Data latency fell from a once-daily batch to an average of 10 minutes, well inside the 15-minute target. Client NPS rose 12 points as proactive, automated alerting replaced the forensic, after-the-fact investigation that used to follow a monthly PDF revealing a problem clients had already noticed themselves.
SLA disputes tied to reporting lag largely disappeared, since clients and the operator's own operations team were now working from the same near-real-time view rather than a client reacting to a month-old summary.
KEEP READING
Explore more studies on AI, analytics, and enterprise intelligence.
