How can factory operations data be monitored and investigated online?
Begin with the decision management needs to make: whether a plan is late, a machine stopped, or a batch has a quality problem. Define the metrics, then integrate ERP/MES, operator reporting, and device data. The first release should move from overview anomaly to line or order, cause and owner, and recorded resolution. That is more valuable than a screen filled with charts.
Online factory visibility is not simply cameras or connecting every machine. The decision needs planned, actual, variance, and response across orders, scheduling, reporting, quality, inventory, equipment, and energy. Misaligned clocks, work-order IDs, and equipment codes make a rapidly refreshed dashboard contradict itself.
When separating device, connectivity, and platform responsibilities, also compare How should a battery management platform be developed? and How should an inventory and production system for a factory be customized?; the linked guidance adds context that should be considered in the same decision.
| Current state | Initial source | First visibility | Limitation |
|---|---|---|---|
| Paper and spreadsheets | Shift forms, scan reporting, inspection entry | Output, completion, defects, downtime reason | Timely and accurate human entry |
| ERP/WMS | API, read-only database sync, document events | Orders, materials, stock, delivery progress | Often lacks operations and live equipment |
| MES/SCADA | Standard interface, message, historian | Orders, operations, equipment, quality | Master definitions and access still need governance |
| Connectable key equipment | PLC/gateway/device API and edge collection | Runtime, stop, count, energy | Retrofit, protocol, security, maintenance cost |
Do not wait until every device connects. Cover one decision loop with people and existing systems, then automate frequent, error-prone, valuable data. Accountable timestamped human reporting can be a sound transition; sensor mappings and clocks can also be wrong.
Every KPI needs a definition, formula and numerator/denominator, time window, source, refresh, owner, and exception treatment. ISO 22400-1 supplies a manufacturing KPI framework but does not decide whether a break or setup is planned downtime for a particular factory. Missing data must appear as missing, not zero or a stale value.
An executive overview can focus on delivery, plan attainment, downtime, quality, and critical stock, but each red value drills into plant, workshop, line, order, machine, or batch with source documents, events, and an owner. The alert creates a work item with acknowledgement, cause, action, and closure. ISA-95 helps assign enterprise, manufacturing, and field responsibilities so the dashboard does not become an unreconciled shadow database or replace PLC safety.
Wavesteam selects a controlled production scope that can demonstrate the chosen decision. Before and during the pilot we measure report effort, latency, anomaly-to-acknowledgement, plan result, and stock variance and sample system values against source records. The observation lasts long enough for the relevant production cycle and anomaly. Expansion follows only when the pilot closes a real issue earlier; persistent data disagreement triggers master-data and process work rather than more charts.