Micro-Stops on the Production Line: How to Detect and Eliminate Them

Micro-stops are the most invisible and most structurally underestimated loss in any industrial production line. Defined as any machine stoppage under five minutes — typically between fifteen seconds and three minutes — they represent on average, across the 450 sites instrumented by TeepTrak in 30 countries, between 8 % and 18 % of a line’s planned production time. Cumulated over the year, that adds up to forty to ninety days of lost production per shift, with no official reporting flagging it.

The paradox of micro-stops is that they are at once small and frequent. Taken individually, none of them warrants the attention of plant management: a 40-second feeder jam, a 90-second sensor fault resolved by the operator, a 2-minute material wait — each is fixed by the operator before the incident log entry could even be completed. Taken collectively, they almost always exceed in cumulative impact the long breakdowns that occupy the front page of the maintenance report. It is this asymmetry between low individual visibility and massive cumulative impact that justifies a dedicated treatment.

This article lays out the complete method for tackling the problem, from initial detection to sustainable elimination. It is written for plant managers, methods engineers, continuous improvement leaders, shift supervisors, and maintenance leaders who recognize a gap between the theoretical performance of their lines and their real performance without being able to quantify it precisely. No proprietary tool is required to apply the method — only adequate measurement instrumentation.

Why micro-stops are invisible at most sites

The primary reason micro-stops are invisible is mechanical: no manual entry system can faithfully capture events that short. When an operator clears a jam in 40 seconds, opening a terminal, selecting a stoppage reason from a dropdown list, entering a comment and validating takes itself 60 to 90 seconds — longer than the incident itself. The operator’s rational arbitrage is not to log it.

Statistically, sites that impose systematic micro-stop logging through procedure achieve at best 40 % capture. The remaining 60 % vanishes into the noise of the daily routine. This under-capture is not an operator discipline problem — it is a measurement-system design problem. As long as a human is asked to manually log events of a few dozen seconds, the loss is inevitable.

The second reason is semantic. The term “micro-stop” itself has no universally normative definition. Depending on the site, the boundary between a micro-stop and a short stop varies between two and ten minutes. This gray zone complicates inter-site comparisons and lets internal discussions slide without converging on a shared definition. International OEE references (JIPM in Japan, AFNOR NF E60-182 in France, SEMI E10 in semiconductors) clearly define the hierarchy of production time but do not set a precise temporal threshold for micro-stops — each site chooses its own.

The third reason is organizational. Micro-stops land in the Performance factor of OEE, not the Availability factor, because they are too short to be logged as stoppages. And the Performance factor is rarely decomposed finely in monthly reviews — people talk about “speed losses” or “slowdowns” without going further. Micro-stops absorb into this generic category and disappear from the steering radar.

Detection: what a sensor sees, what a manual log misses

Automatic micro-stop detection rests on continuous machine-state measurement, independent of the operator. Three families of instrumentation coexist on the industrial market in May 2026.

The first is direct PLC integration: capturing the machine’s “in production” bit via OPC-UA, Modbus, or proprietary protocol. The advantage is absolute precision — the PLC knows exactly when the machine is running. The disadvantages are operational: PLC access locked by IT/OT, mandatory change control in pharma and food (typically six to twelve weeks of validation), frequent incompatibility with older PLCs (before 2010) that lack standard network interfaces, and a need to adapt PLC code on certain architectures. Hidden engineering integration hours are substantial.

The second is MES event capture: pulling stoppages from the existing MES. The advantage is that no hardware installation is required if the MES is already deployed. The major disadvantage is that the MES captures only what operators declare — so it captures micro-stops no better than manual entry. It is a false remedy to the visibility problem.

The third is wireless external sensor instrumentation, where TeepTrak has been a French pioneer since 2015. Autonomous IoT sensors (vibration, current, optical, accelerometer) are installed on the equipment surface in under thirty minutes per machine, with no PLC modification, no change control, no IT validation. Detection rests on the physical signature of the machine state rather than a logical bit. Precision reaches two to five seconds depending on configuration. It is the approach with the best ratio of deployment speed to detection precision in May 2026 on the majority of industrial sites.

Diagnosis: from raw detection to root causes

Detecting micro-stops is not enough — you also need to understand what triggers them in order to act. Moving from raw data to root causes follows four sequential steps on the field.

Step 1 is the qualification of each stoppage by category. Once automatic detection is in place, operators or shift supervisors qualify each recorded stoppage from a touch terminal, selecting a cause from a list parameterized for the line. Unlike full manual entry, the time is already measured — the operator only has to label, which takes two to five seconds per event. The qualification rate typically reaches 85 to 95 % after four weeks of routine.

Step 2 is Pareto sorting over four to six weeks of data. Once qualification stabilizes, the classic Pareto analysis surfaces the three to five reasons accounting for 70 to 80 % of the cumulative micro-stop volume. On nearly every line measured by TeepTrak, these three to five reasons are already known to the field team — but their relative weights almost always surprise. Numbers-based prioritization replaces intuition-based prioritization.

Step 3 is the causal diagnosis on the dominant reasons. Classic Lean tools apply: 5 Whys, Ishikawa diagram, Failure Mode and Effects Analysis (FMEA). The temporal granularity brought by automatic detection changes the game — you can now correlate micro-stop peaks with external factors: changeovers, time slots, shifts, material lots, ambient conditions. The correlation often reveals an external cause that no field team had suspected.

Step 4 is the prioritization of projects. Each reason identified as a root cause is evaluated on two criteria: impact (recoverable OEE points) and feasibility (implementation effort). The standard impact-feasibility matrix surfaces two to three priority projects for the next six weeks. The continuous improvement team focuses its effort on those projects.

Elimination: the three families of projects that work

Experience accumulated across 450 sites shows that micro-stop elimination projects fall into three families whose success rates and timelines differ significantly.

The SMED-applied-to-micro-stops family tackles stoppages tied to changeovers, adjustments, and calibrations. The standard SMED method (Single-Minute Exchange of Die), developed by Shigeo Shingo in the 1970s, remains the reference approach fifty years later. On the sites measured, it typically reduces average changeover time by 40 to 60 %, with a substantial portion coming from eliminating peri-changeover micro-stops (failed first piece after changeover, quality validation wait, post-restart adjustment). Typical implementation timeline: eight to twelve weeks per line.

The workstation standardization family tackles stoppages tied to operator gestures: repeated adjustments, manual verifications, walks to fetch a tool. The reference tool is the standard work sheet (SWS) or Standard Work, complemented by a workstation rearrangement following 5S principles. Observed gains are 25 to 40 % on micro-stops in this category, with an implementation timeline of four to six weeks.

The preventive and conditional maintenance family tackles stoppages tied to gradual equipment degradation: fouling, mechanical play, sensor drift, lubrication degradation. Measurement granularity allows detection of early degradation signatures and triggering of preventive intervention before the micro-stop. Gains take longer to materialize — six to twelve months — but are sustainably structural. It is the family where TeepTrak observes the most enduring return on instrumentation.

Measuring the return: quantifying micro-stop gains in OEE points

Quantifying the return of a micro-stop program requires a rigorous baseline measurement, otherwise any result is contestable. The standard method consists of measuring for four to six weeks before the project starts (baseline), then for the same duration after corrective actions, under comparable production conditions (same products, same shifts, same seasons).

Typical gains observed by TeepTrak on sites that apply the complete detection + diagnosis + action method are the following. Over 12 months, the reduction in micro-stop volume reaches 35 to 50 % versus baseline, translating into an OEE gain of 6 to 12 points depending on the initial dominance of the Performance factor in the OEE calculation. Over 24 months, the most disciplined sites reach 60 to 70 % reduction, i.e., a cumulative OEE gain of 10 to 18 points.

These gains are mechanically superior to those obtained by actions targeted at long breakdowns, because cumulative micro-stop volume is larger at the starting point. This asymmetry is what justifies prioritizing a micro-stop project before a classic preventive maintenance project on most production lines in May 2026.

For detailed OEE calculation and per-factor decomposition, the dedicated article How to Calculate OEE: Formula, Method, and Worked Example covers the formula and applies it on a complete worked example.

Common mistakes that derail a micro-stop program

Five mistakes systematically appear in micro-stop programs that miss their targets, regardless of sector or site size.

• Launching without a measured baseline. If the initial micro-stop volume measurement rests on manual entry, the baseline structurally understates the problem — and real gains are incomparable to any historical data. Automatic detection must precede any action, not accompany it.
• Mixing micro-stops and breakdowns. The two categories call for radically different levers. A steering report that aggregates them into a single indicator blurs the diagnosis and misroutes the projects.
• Asking the operator to quantify the micro-stop. The operator qualifies (labels the cause) but does not quantify (duration measured by the sensor). Any confusion on this role split kills capture rate within four weeks.
• Skipping the Pareto phase and attacking the “most visible” reason. Intuition almost always points to the most memorable reason, not the most impactful. On half of sites, the “most visible” reason is not even in the real Pareto top three.
• Failing to sustain measurement after the project. Without ongoing measurement, the gain achieved degrades within six to twelve months through return to prior behaviors. Measurement must remain in place after the project, not just during.

For depth on Pareto + targeted SMED elimination, see the satellite article Eliminating Micro-Stops: Pareto, Top 3 Root Causes, and Targeted SMED Project. For pure technical detail on sensor detection, see Detecting Micro-Stops: External Sensors, Thresholds, and False Positives.

Typical timeline of a micro-stop program on one line

A complete micro-stop program on a single line, from decision to first measured gains, typically deploys over sixteen to twenty weeks in May 2026. The standard sequence is the following.

• Weeks 1-2: sensor installation and calibration. Physical mounting, detection threshold parameterization, stoppage cause configuration in the qualification terminal. No gain expected at this stage.
• Weeks 3-8: baseline and qualification. Six weeks of stabilized measurement during which the field team qualifies every stoppage and the Pareto analysis is built. By the end of this phase, the three to five dominant reasons are identified.
• Weeks 9-12: first targeted project. Selection of Pareto reason #1, causal diagnosis (5 Whys, Ishikawa), corrective action (SMED, standardization, or preventive maintenance depending on nature). Gain measured in parallel.
• Weeks 13-16: second targeted project and consolidation. Project launch on Pareto reason #2, audit of first project sustainability. First OEE outcome review at sixteen weeks.
• Weeks 17-24 and beyond: continuous improvement routine. A new reason is addressed every two to three months. Measurement stays in place. Line OEE follows an upward trajectory measurable month over month.

This is the standard sequence observed at the most rigorous TeepTrak client sites. The main source of timeline drift is not the technical difficulty of the projects, but the field-team capacity to conduct them — a micro-stop program is demanding in shift-supervisor and continuous-improvement-engineer hours during the first six months.

Detect and eliminate your micro-stops — 48 hours, no PLC modificationWireless external sensors installed in under 30 minutes per machine. Automatic micro-stop detection from 2 seconds. Free 48-hour POC on site, 450+ factories deployed across 30 countries.Request an OEE demo

External references

OEE — Wikipedia · JIPM — Japan Institute of Plant Maintenance · SME — Society of Manufacturing Engineers · AFNOR — NF E60-182

Related TeepTrak reading: Detecting Micro-Stops: External Sensors, Thresholds, and False Positives · Eliminating Micro-Stops: Pareto, Top 3 Root Causes, and Targeted SMED Project · The Hidden Cost of Micro-Stops Your MES Does Not See · How to Calculate OEE: Formula, Method, and Worked Example