Where utility monitoring programmes stall — and what each needs.
Water utilities rarely lack data; they lack data anyone can act on before the loss is already banked. These are the patterns we see across municipal and bulk networks.
| What goes wrong | How you notice | The fix |
|---|---|---|
| Meters read monthly, by walk-past | A problem is discovered weeks after it started | Automate the read. A monthly figure cannot separate a leak from a billing question |
| Telemetry without a maintenance route | Nodes drop off one by one and the network quietly shrinks | Treat node health as an asset register with an owner and a schedule, not as an IT afterthought |
| Alarms routed to a shared inbox | Everyone assumes somebody else responded | Route to a person on duty with an escalation path, and log the response beside the alarm |
| Reservoir level without pump status | A dry-run or a failed pump looks like ordinary demand | Pair level with pump state or current; the two together are a diagnosis, level alone is a symptom |
| Data gaps through power outages | The record fails on exactly the days the network was stressed | Buffer at the node. A network measured only while the grid is up is measured on its easiest days |
| No baseline before intervention | Improvements are argued rather than demonstrated | Take a clean period of data first — without it, every recovery claim is contestable |
| Compliance reporting separated from operations | The same readings are collected twice, for two audiences | Collect once and report twice. Operational data and the compliance record should come from the same source |
The two that cost most are orphaned telemetry and the shared inbox. Nodes installed without an owner degrade at a predictable rate, and the coverage map on the wall stops matching the network within a year. Alarms sent to an address rather than to a person fail the same way: nothing is ignored deliberately, and nothing is answered either.