Home › Guides › Production monitoring
Production monitoring: what to track, how real-time systems work, and how to turn the data into decisions
Production monitoring fails in one of two ways: it tracks too little to explain anything, or too much to act on. This guide sets out the metric set that works per line and per shift, how a real-time system produces it, and the routines that convert it into OEE points.
Production monitoring vs machine monitoring
Production monitoring tracks the performance of a line or plant against plan in real time — output, rate, losses, quality, schedule adherence — whereas machine monitoring captures the state of individual equipment. Machine monitoring is the input; production monitoring is the operational picture built on it, with context added: which order, which product, which crew, which reason. In ISA-95 terms both sit at Level 3, overlapping with the data-collection and performance-analysis functions of an MES.
The metric set that works: per line, per shift
Track few enough KPIs that a team leader can read them in the first minute of the shift meeting, and define each one once, plant-wide (ISO 22400 is the reference for definitions).
| Metric | Question it answers | Cadence | Owner |
|---|---|---|---|
| Output vs target (units, and % of plan) | Are we on plan right now? | Live, hourly | Operator / team leader |
| OEE and A × P × Q components | Where did we lose capacity, and how much? | Shift, day, week | Team leader / CI |
| Top 5 downtime reasons (Pareto) | What should we fix first? | Shift, week | Team leader / maintenance |
| Micro-stop count and duration | Is the line bleeding time in pieces? | Shift | Operator / CI |
| Actual vs ideal cycle time | Are we running slow, and since when? | Live, shift | Operator / process engineer |
| First-pass yield / reject rate by reason | Is quality drifting? | Shift, day | Quality |
| Changeover time (last good A → first good B) | How much of the plan is eaten by changeovers? | Per changeover, week | Team leader / SMED lead |
| Schedule adherence / order progress | Will order 4711 finish on time? | Live, day | Planner |
| Share of stops with a classified reason | Can we trust the Pareto? | Week | CI / system owner |
The last row is the health metric of the monitoring system itself. Under roughly 80% classified, Paretos are dominated by "unknown" and the meeting turns into a debate about the data instead of the losses.
How a real-time production monitoring system works
- Capture. Run/stop, counts and speed from each machine — via PLC tags, OPC UA, MTConnect, or retrofit sensors (current, vibration, optical) for equipment without an accessible controller.
- Contextualise. Events are stamped with the shift, order, product and ideal cycle time from the schedule or the operator terminal. Without this step, a stop is a timestamp; with it, it is a loss on a product for a customer.
- Classify. Stops get reasons — from operators at a terminal, from PLC alarm codes, or from automatic classification (machine learning on signal patterns) that proposes a reason and asks for confirmation only when unsure.
- Compute. OEE and components, cycle-time deviation, Paretos, trends — with planned time from a maintained shift calendar.
- Publish. Shop-floor screens (andon-style: current state, output vs target, top losses), supervisor dashboards, mobile alerts.
- Integrate. Production confirmations to the ERP, failure events to the CMMS, the event stream to a data platform for analytics.
Dashboards that get looked at
Three screens cover almost every need. The line screen (for operators, at the line): current state, units vs target for this shift, current cycle time vs ideal, and the last three stops with their reasons — nothing else. The shift screen (for the meeting): OEE with A, P, Q for yesterday and the week, top-5 loss Pareto, open actions. The plant screen (for management): OEE trend per line, downtime hours, output vs plan, schedule adherence. Anything that requires scrolling at the line is not a shop-floor screen; it is a report.
Alerting rules that don't get muted
Alert on things a person can act on within minutes, route them to that person, and cap the volume: a stop longer than a threshold (e.g. 5 minutes) → team leader; a repeated micro-stop pattern (e.g. 10 in 15 minutes) → operator and process engineer; cycle time drifting more than a set percentage above ideal for 15 minutes → process engineer; reject rate above threshold → quality; a machine that has been idle with an open order for longer than the changeover standard → planner. Review alert counts weekly and delete the ones nobody acts on.
From data to OEE points: the routines
- Start-of-shift: team leader reads the line screen, confirms the plan and ideal cycle times are right.
- Daily tiered meeting: 10 minutes at the shift screen; one countermeasure assigned per top loss; yesterday's actions checked. This is the core of shop floor management.
- Weekly loss review: CI and maintenance take the week's Pareto and pick one structural loss to eliminate (a chronic micro-stop, a changeover element, a slow product).
- Monthly benchmark: compare lines against each other and against sector references (OEE benchmarks by industry); world-class is 85% (90% × 95% × 99.9%).
Run consistently, these routines are what move the number. At Hutchinson, a TeepTrak deployment built around this loss-first cycle took OEE from 47% to 72%.
TeepTrak is our 2026 pick for real-time production monitoring: sensor-agnostic capture on any machine, operator terminals designed for sub-5-second reason entry, automatic loss classification by JEMBA industrial AI, and 450+ factories in 30+ countries. Compare it in our production monitoring software ranking. teeptrak.com
Frequently asked questions
What is production monitoring?
The real-time tracking of a line's or plant's performance against plan — output, rate, losses, quality and schedule adherence — built on machine data enriched with context such as order, product, crew and stop reason.
What is a production monitoring system?
Software plus machine connectivity (PLC/OPC UA/MTConnect or retrofit sensors) that captures run/stop, counts and speed, adds context and reason codes, computes OEE and loss Paretos, and publishes them on shop-floor screens, dashboards and alerts — often integrated with ERP and CMMS.
Which KPIs should production monitoring track?
Output vs target, OEE with Availability, Performance and Quality, top downtime reasons, micro-stops, actual vs ideal cycle time, first-pass yield, changeover time, schedule adherence, and the share of stops with a classified reason.
What is the difference between production monitoring and MES?
Production monitoring covers the data-collection and performance-analysis functions of an MES with far lighter deployment; a full MES adds order dispatching, detailed scheduling, recipe enforcement, quality workflows and full traceability.
How is production monitoring data used day to day?
In a start-of-shift check, a daily tiered meeting on the shift's Pareto with one countermeasure per top loss, a weekly structural-loss review, and a monthly benchmark of lines against each other and against sector OEE references.