Autonomous Maintenance: A Practitioner's Guide to Building Operators Who Own Their Equipment
- May 2
- 33 min read
Updated: 10 hours ago
By Allan Ung | Founder & Principal Consultant, Operational Excellence Consulting (OEC)
Updated: 15 May 2026
Allan Ung is the Founder and Principal Consultant of Operational Excellence Consulting (OEC), a Singapore-based management consultancy established in 2009. With over 30 years of experience leading operational excellence and quality transformation across manufacturing, technology, and global operations — including senior roles at IBM, Microsoft, and Underwriters Laboratories (UL) across Asia-Pacific — Allan brings deep shopfloor and strategic expertise to every engagement. He holds the following qualifications and recognitions: Certified Management Consultant (CMC, Japan), Certified Lean Six Sigma Black Belt, JIPM-certified TPM Instructor, TWI Master Trainer, and former National Examiner for the Singapore Business Excellence Award. Allan has designed and facilitated TPM implementations and operational excellence programmes for organisations across semiconductor, automotive, and industrial manufacturing sectors. His clients include Temic Automotive (Continental), Analog Devices, Amkor Technology, STATS ChipPAC, Panasonic, Micron, Lam Research, Infineon Technologies, Dorma, and Tokyo Electron.
The Question That Reveals Everything About a Plant's Maintenance Culture
Walk onto a shopfloor and ask a machine operator what they do when they notice an unusual sound coming from their equipment. The answer to that question — more than any OEE figure, more than any maintenance KPI dashboard — will tell you almost everything about the state of equipment ownership in that organisation.
In plants with a conventional maintenance culture, the answer is usually some version of: "I call maintenance and wait." In some cases, the operator has learned through experience that raising an alert too early leads to a dismissive response from an overloaded maintenance team, so they wait until something more definitive happens — meaning they wait until a breakdown occurs. The equipment fails, production stops, and everyone treats it as an unforeseeable event, when in fact the signals had been present for days, even weeks, and a trained set of eyes could have caught them.
In plants where Autonomous Maintenance has taken root, the answer is different. Operators describe a moment of pattern recognition — they know this machine's normal operating sounds well enough to distinguish a change, they know what it likely means, they know whether they can address it immediately or need to tag it for the maintenance team, and they know the exact protocol for doing so. The machine is, in a meaningful sense, theirs. They are not operating it; they are caring for it.
That difference — between a culture of passive operation and one of active ownership — is what Autonomous Maintenance is designed to create. Having implemented AM programmes across semiconductor fabrication facilities, automotive assembly plants, precision engineering workshops, and industrial manufacturing operations across Asia-Pacific over more than three decades, I can tell you with confidence that AM is simultaneously the most powerful and the most misunderstood pillar of Total Productive Maintenance. This guide is an attempt to address that gap — not with theory, but with the practitioner's view of what AM actually requires to work.
What Autonomous Maintenance Actually Means — and What It Does Not
Autonomous Maintenance, as defined within the JIPM framework of Total Productive Maintenance, refers to TPM activities that involve operators in maintaining their own equipment, independent of the maintenance department. The goals are threefold: to prevent equipment deterioration, to restore equipment to its ideal state, and to establish the basic conditions needed to keep equipment well maintained on an ongoing basis.
That definition sounds straightforward. In practice, it is routinely misread. I have encountered organisations that describe their AM programme as operators performing cleaning during shift changeover, with no structured inspection, no tagging protocol, no standards, and no learning built into the activity. That is not Autonomous Maintenance — it is cleaning duty dressed up in TPM terminology. I have also seen operations where management introduced AM primarily as a way to reduce the maintenance department's workload during a period of cost pressure, with operators handed a checklist and told to get on with it. That approach almost always fails, and when it does, it poisons the well for any future AM effort.
AM is not a favour to the maintenance department. It is not an activity to keep operators busy when production demand is low. It is not about turning operators into skilled tradesmen, and it is not a one-time exercise. These misconceptions matter because they shape how the programme is launched, resourced, and sustained — and getting those fundamentals wrong is the single most common reason competent organisations invest in AM and fail to see results.
At its core, AM rests on a paradigm shift in the relationship between operators and their machines. The old attitude, still entrenched in many organisations, draws a hard line: operators run equipment, maintenance fixes it. TPM's fundamental challenge to that model is captured simply: what if the people closest to the machine every day — who see it, hear it, and feel it for eight or twelve hours a shift — were also equipped and empowered to be its primary caregivers? What would change if operators moved from "I use" to "I maintain and I improve"?
The answer, consistently demonstrated across the facilities I have worked with, is that breakdowns decline, defects fall, OEE rises, and operators themselves develop a pride in their equipment and their work that no engagement survey can manufacture artificially.
The Iceberg Beneath the Breakdown
One of the most important conceptual shifts AM requires is a change in how operators and managers think about machine failures. The common view of a breakdown is an event — the machine stops, production halts, maintenance is called. The underlying view that TPM instills is structural: a breakdown is never an isolated event. It is the visible tip of an iceberg of minor defects, accumulated contamination, loosened fasteners, inadequate lubrication, and ignored abnormalities that have been building beneath the surface, often for months.
Minor machine defects are generally unnoticed but are the cause of almost all machine failures. Contamination leads to corrosion; corrosion leads to leaks; leaks lead to deformation; deformation leads to vibration; vibration leads to loosening; loosening leads to failure. The causal chain is rarely mysterious after the fact — it is only invisible beforehand because nobody was looking at the right level of granularity, with the right knowledge, consistently enough to catch the early signals.
The bathtub curve from reliability engineering makes this visible in a different way. Every piece of equipment passes through three phases: infant mortality failures in the early period of use, a period of relatively stable, constant failure rate in the useful life phase, and an accelerating wear-out failure rate as components reach the end of their design life. Preventive maintenance — and Autonomous Maintenance specifically — works by extending the useful life phase. AM addresses infant mortality by ensuring equipment is correctly cleaned, lubricated, and inspected from the outset. It addresses the wear-out phase by detecting deterioration early, before it cascades into failure.
It is also worth distinguishing between natural and forced deterioration, a distinction that sits at the heart of AM's philosophy. Natural deterioration is the normal wear that occurs when parts rub against one another — it is the physical reality of mechanical operation, and it proceeds at a predictable rate. Forced deterioration is something different: deterioration that happens sooner than it would naturally, typically caused by failures to maintain basic conditions. Not keeping parts clean and lubricated, ignoring excessive loads in moving parts, allowing contamination to accumulate unchecked — these are the acts of omission that accelerate wear and create breakdowns that look random but are entirely preventable. AM's primary aim is to eliminate forced deterioration entirely.
The Four Equipment-Related Skills Operators Must Develop
Before going into the seven-step AM implementation sequence, it is worth establishing what AM is actually trying to build in the people who run equipment. JIPM's framework identifies four equipment-related skills that operators must progressively develop through the AM journey.
The first is the ability to detect abnormalities — to notice that something about the machine's condition, sound, smell, vibration, or appearance has changed from its normal state. This skill sounds basic but it requires a baseline: operators can only detect abnormalities if they first understand what normality looks like. Initial cleaning, which is Step 1 of AM, creates that baseline by giving operators direct, hands-on contact with every surface and component of their machine.
The second skill is the ability to correct and restore abnormalities — to act on what they detect, either by performing an immediate fix within their authorised scope or by tagging the problem for maintenance intervention. This requires both technical knowledge and a clear protocol that removes ambiguity about who does what.
The third skill is the ability to set and maintain optimum equipment conditions — to understand not just that a machine is or is not working, but what conditions are required for it to operate at its best, and to actively maintain those conditions through cleaning, lubrication, tightening, and inspection.
The fourth skill, which only emerges after the first three are established, is the ability to manage equipment autonomously — to exercise ongoing, self-directed care of the machine as part of daily work, supported by standards, visual controls, and a continuous improvement mindset. This is the skill that characterises a mature AM culture, and it typically takes two to four years of structured programme implementation to develop reliably.
The Seven Steps of Autonomous Maintenance: A Practitioner's Walkthrough
The seven-step AM framework, adapted from the Japan Institute of Plant Maintenance, is a carefully sequenced development pathway that builds operator capability progressively. Understanding the logic of the sequence is as important as understanding what each step requires — because the steps are not interchangeable, and skipping or rushing any of them undermines the foundation that the subsequent steps depend on.
The sequence is deliberate: Steps 1–3 establish basic equipment conditions; Step 4 builds the technical knowledge to inspect intelligently; Steps 5–7 embed autonomous care as a sustainable, self-managed discipline. Each step has an audit gate — teams do not advance until the current step's capability has been genuinely built.
Step 1: Clean and Inspect
The first step is the most revealing and, in my experience, the most underestimated. Its purpose is to eliminate all dirt and grime from the machine, lubricate and tighten bolts, and find and correct problems. In practice, it involves a thorough, hands-on initial cleaning of the equipment by the operators themselves — often for the first time in the machine's operating history in that facility.
What makes this step foundational is the principle that cleaning is inspection. When an operator wipes down a machine by hand rather than leaving it to be periodically jet-washed, they make contact with every surface. They feel the warmth of a component that should be cool. They notice the vibration in a section of the machine they had never paid attention to. They find the bolt that has been missing for three months, the oil that has been slowly seeping from a joint, the crack in a casing that is just beginning to propagate. None of these would show up on a breakdown report, because none of them have caused a breakdown yet. But left unaddressed, all of them will.
The tagging system introduced in Step 1 is the practical mechanism that makes discoveries actionable. When an operator finds a problem during initial cleaning, they attach a physical tag to it — a blue tag if it is something within the operator's scope to address, a red tag if it requires maintenance department involvement. The contents of each tag are transferred to a tag register, and resolution is tracked until every tag has been cleared. This visible, tangible record of discovery and resolution is enormously motivating for teams experiencing AM for the first time. In one semiconductor assembly facility I worked with, the initial cleaning of a single press revealed forty-seven tagged abnormalities. Nineteen were cleared within the first week. The operators who had run that machine for years had no idea those deficiencies existed.
In an OEE benchmarking study I facilitated in 2014 across three semiconductor manufacturers in the Philippines — Analog Devices General Trias (ADGT), STATS ChipPAC, and Amkor Technology Philippines — the variation in AM practice between organisations with otherwise comparable OEE results was striking. ADGT conducted daily equipment AM during lot inspection and tracking at the start of each shift, integrating the activity into the shift transition rhythm. Operators at STATS ChipPAC used a formal checksheet for daily AM, with some machines opened by operators to clean contactors inside the equipment — a level of hands-on access that reflects genuine ownership. At Amkor, cleaning was performed at the start of shift, with music used as an auditory cue to make the activity part of the shift's opening discipline. Three organisations, three sets of machines carrying comparable OEE figures, and three meaningfully different levels of structured AM practice. What this comparison made visible is that OEE parity at a given moment does not indicate AM maturity parity — it indicates that some organisations have compensated for weaker AM with other factors, and that the advantage of structured AM shows up most clearly over a longer time horizon, as basic equipment condition either holds or degrades. Equipment age was illustrative: ADGT's UltraFlex equipment averaged just two years old, while the equivalent platforms at STATS averaged eight years and at Amkor seven. That organisations with significantly older equipment were sustaining comparable OEE figures speaks to the power of consistent operator care — and raises the question of what OEE trajectories would look like once that equipment ages further without a structured AM programme to slow forced deterioration.
Supervisors and technicians should agree in advance on a date for Step 1 implementation, identify all materials needed, arrange for maintenance support on the day, and ensure that operators are coached to look beyond the obvious surfaces — into recesses, under covers and lids, on auxiliary equipment like conveyors and gauges and electrical cabinets. The principle is clear: thoroughly clean the chosen machine and never leave the job half done. A partial initial cleaning produces a partial baseline, which means the foundation for all subsequent steps is compromised.
Step 2: Eliminate Problem Sources and Inaccessible Areas
Once initial cleaning is complete and abnormalities have been tagged and progressively resolved, Step 2 addresses a category of problem that cleaning alone cannot solve: the design and environmental conditions that make the machine dirty in the first place. The focus is on correcting sources of dirt and grime, preventing spattering, improving accessibility for cleaning and lubrication, and shortening the time it takes to perform basic maintenance tasks.
This step is where TPM begins to involve engineering problem-solving. Countermeasures to contamination sources — using acrylic covers to contain cutting oil mist, installing hinge doors rather than fastening screws to improve access, redesigning guards to minimise scatter while maintaining safety — require creativity and technical collaboration between operators and engineers. Where-where analysis traces the path of contamination to its origin; why-why analysis uncovers the root cause. Both tools are used systematically in Step 2 to ensure countermeasures address causes, not symptoms.
The accessibility dimension of Step 2 is particularly important, and it is one that is frequently neglected. If cleaning a particular section of a machine requires thirty minutes because the access panels are designed for maintenance overhaul rather than daily operation, operators will simply not clean that section consistently under normal production pressure. The goal is to make the equipment easier to clean and easier to inspect — installing inspection windows, simplifying wiring and piping layouts, repositioning lubrication inlets, changing lubrication methods to reduce time requirements. Every improvement that reduces the time and effort required for basic maintenance directly increases the probability that the standard will be followed.
Step 3: Develop Cleaning and Lubricating Standards
The insights gained from Steps 1 and 2 are only valuable if they are codified into standards that can be followed reliably by all operators, across all shifts, consistently over time. Step 3 is where that codification happens. The output is a set of standards covering what needs to be cleaned, inspected, and lubricated; how each task should be performed; how frequently; and by whom.
The significance of lubrication standards deserves particular emphasis. Data from the Japan Institute of Plant Maintenance points to a striking figure: approximately 60% of breakdowns in the moving parts of machine tools are attributable to poor lubrication or oiling. In plants I have assessed across Asia-Pacific, the three root causes are consistent: the absence of lubrication standards, the failure to adhere to existing standards, and the use of incorrect lubricants or methods. All three are preventable with well-designed, well-communicated standards.
Effective lubrication standards specify the lubrication inlet locations and how to verify them, the correct lubricant type for each point, the lubrication method and tools to be used, the quantity and frequency of application, and the person responsible. They are based on actual operating experience and the equipment manufacturer's manual, and they are realistic — standards that cannot be met within the allocated time budget will simply be ignored. The TWI Job Instruction format, with its structure of major steps, key points, and reasons for key points, is particularly effective for capturing these standards in a form that operators can be trained on and that new team members can learn from.
A lubrication control system, developed with the maintenance department, brings all of this together: identifying lubricating points and surfaces, allocating routine lubrication tasks between operators and maintenance technicians, drawing lubrication system flowcharts, and implementing a colour-coded system that removes any ambiguity about which lubricant goes where. Colour-coded oil cans matched to colour-coded lubrication point labels eliminate the most common source of lubrication error: using the wrong product because the labelling was unclear.
Step 4: Conduct General Inspection Skills Training
Steps 1 through 3 address what operators do; Step 4 addresses what they know. At this point in the AM journey, operators are cleaning, inspecting, and lubricating their machines according to documented standards. But for those activities to be genuinely effective — for operators to catch the early warning signs that prevent failures rather than simply going through the motions of a checklist — they need to understand how their equipment actually works.
Step 4 involves structured skills training, typically organised by equipment subsystem: pneumatics, hydraulics, electrical systems, drives, lubricants and coolants, fasteners, and safety devices. The training is conducted using inspection manuals developed with the maintenance department and is designed to give operators the technical vocabulary and conceptual understanding to use their senses — sight, hearing, smell, and touch — as diagnostic instruments. Abnormal hydraulic pump sounds can indicate cavitation or pump failure. Unusual heat in a drive component signals bearing wear. An unexpected smell near an electrical cabinet may precede a fault. These diagnostic connections, which an experienced maintenance technician makes automatically, must be taught explicitly to operators.
Visual controls are a central tool in Step 4, and their design deserves careful thought. A visual control is effective only if it makes the normal condition immediately distinguishable from an abnormal one at a glance, without requiring specialised knowledge to interpret. A gauge label with green and red zones tells any operator immediately whether the reading is acceptable. Match marks on fasteners — painted lines that cross from the bolt head onto the surrounding surface — make it instantly visible whether a bolt has rotated from its torqued position. Colour-coded lubrication inlets ensure the right lubricant is used without the operator needing to consult a reference document. The design principle is consistent: make the abnormal condition impossible to miss and impossible to misinterpret.
One-point lessons are the training vehicle that brings Step 4's knowledge-sharing to life at the team level. Each one-point lesson covers a single aspect of equipment structure, function, or maintenance practice on a single sheet of paper, presented in five to ten minutes, using more visual information than text. Critically, one-point lessons are developed and taught by operators themselves, not delivered by trainers. An operator who discovers that a particular pneumatic fitting is prone to loosening at a specific temperature range writes a one-point lesson about it, explains it to teammates at the next standup meeting, and the knowledge transfers directly to the people who need it. This cascading peer-teaching model builds ownership and expertise simultaneously.
The skills matrix board, updated as operators complete training and demonstrate competency in each AM task, makes capability visible at the team level and identifies gaps that require attention. In plants where AM is genuinely embedded, skills matrix boards are live documents actively maintained by team leaders — not laminated placards on the wall that have not been updated in two years.
Step 5: Conduct Autonomous Inspections
By Step 5, operators have cleaned and restored their machines, eliminated contamination sources, developed maintenance standards, and built the technical knowledge to inspect intelligently. The focus now shifts to embedding autonomous inspection as a reliable daily practice. Operators prepare standard check sheets for autonomous inspection, define an inspection schedule for each individual, carry out the inspections consistently, and refine their methods based on accumulating experience.
The process of designing effective autonomous inspection check sheets requires careful collaboration between operators, supervisors, and the maintenance department. The check sheet should cover what to inspect, using which method, at what frequency, with what criterion for normality, and what action to take when a condition is abnormal. The language should be concrete and unambiguous — not "check bearing temperature" but "confirm bearing surface temperature is below 60°C using the temperature probe; tag for maintenance if above threshold."
A critical discipline at Step 5 is the formal coordination of activities between autonomous maintenance and specialised maintenance. By the end of Step 5, there should be clear, mutually agreed standards for what operators maintain and what maintenance technicians maintain; a reviewed and refined set of tentative inspection standards based on operating experience; a yearly maintenance calendar; and a complementary set of standards for major equipment servicing performed by the maintenance department. Getting this coordination right eliminates both the overlap (where both teams think the other is handling something) and the gaps (where neither team thinks they are responsible).
Over time, Step 5 also involves a refinement of the inspection list itself. Based on accumulated experience, items that have never revealed a problem after many cycles of inspection may be removed or their frequency reduced. Items that have consistently surfaced early warnings of developing problems are retained and potentially intensified. This is a data-driven, experience-guided evolution of the standard — not an arbitrary simplification.
Step 6: Standardize through Visual Workplace Management
Steps 1 through 5 focus primarily on individual machines and equipment-centred tasks. Step 6 widens the lens to the entire work area, extending the principles of standardisation and visual management to all work processes. The standards required at this step encompass cleaning, lubrication, and inspection standards for the broader work environment; shopfloor materials flow standards; data recording method standards; and tool and die management standards.
Visual route maps for frequent preventive maintenance tasks are a practical tool at this stage. In facilities running multiple shifts, colour-coded route maps — with shape indicating inspection frequency, colour indicating shift ownership, and numbers indicating sequence — allow any operator to execute the correct inspection route without ambiguity, even on shifts when the usual team member is absent. Agilent's IC fabrication facility has used this approach effectively, and I have adapted it for several semiconductor assembly operations in Singapore and Philippines.
The broader 5S workplace organisation that Step 6 reinforces is not a separate initiative running parallel to AM — it is an integral part of AM's maturation. An organised workplace where every item has a designated location, where necessary spares are planned and stocked, where work instructions are consistently followed, and where process quality is actively assured through daily checks is the environment in which autonomous maintenance actually becomes sustainable. Without it, the gains from Steps 1 through 5 gradually erode as the workplace returns to its previous state.
Step 7: Implement Autonomous Equipment Management
Step 7 is not a new set of activities — it is the institutionalisation of everything that has been built through the previous six steps into the normal operation of the business. Company policies and objectives for autonomous equipment management are formalised. MTBF (mean time between failures), MTTR (mean time to repair), and MTTF (mean time to failure) data are collected reliably, analysed consistently, and used to drive equipment improvement decisions. Improvement activities — kaizen — become a standard part of daily work, not a special event.
The OEE targets and other TPM metrics that the organisation is pursuing are achieved at Step 7 not through occasional bursts of effort but through three interlocking activities: maintaining what has been established, improving on it continuously, and passing the knowledge on to new team members so that the accumulated capability does not erode with personnel turnover. The PDCA cycle — Plan, Do, Check, Act — is the management rhythm that sustains this.
At Step 7, management's role shifts from implementation support to cultural reinforcement. Department heads conducting regular shopfloor inspections — monthly at a minimum — are not checking on compliance; they are communicating priorities, providing guidance, and assessing whether team activities are on track. Division-level inspections on a quarterly basis reinforce that AM is a strategic commitment, not a production department initiative. When senior leaders visibly engage with AM activity boards, ask intelligent questions about tag resolution rates and OPL counts, and connect AM performance to business outcomes, the cultural signal is unmistakable. When they do not, the message is equally clear.
The Three Key Tools That Make AM Teams Function
Three practical tools give AM teams their operating rhythm: activity boards, daily standup meetings, and one-point lessons. Each one is worth examining in enough depth to understand why it matters.
The activity board is a visual management tool that makes the team's AM journey visible to everyone — the team itself, supervisors, management, and other teams who visit the area. Effective activity boards display the team's strategy and current priorities, progress against each AM step, statistical trends for OEE and the six big losses, key issues and the actions being taken to address them, a record of abnormalities discovered, case studies of improvements, and one-point lessons. The board is updated by hand, by the team. A laminated board managed by a coordinator is not an activity board — it is a display case. The difference between a living tool and a static artefact is whether the team members who run the equipment are the ones maintaining it.
Daily standup meetings — short, structured, held at the activity board — are where the board's content is activated. The agenda covers yesterday's issues and lessons learned, manpower status for the shift, updates from management, Lean and TPM activities, and today's targets and actions. The discipline is brevity and focus. Meetings that become complaint sessions or status reports without action items lose their value quickly. Meetings that are reliably short, consistently structured, and genuinely productive become a tool that teams protect.
One-point lessons are the knowledge management infrastructure of AM, and their value compounds over time. A team that has been running AM for three years and has accumulated one hundred and fifty one-point lessons — covering basic knowledge about equipment subsystems, examples of problems encountered and resolved, and examples of improvements made — has created a body of knowledge that exists nowhere else: it is drawn from their specific machines, their specific operating conditions, their specific failure modes. When that knowledge is cascaded through peer teaching rather than stored in a filing cabinet, it becomes a living capability rather than a documentation artefact.
Where AM Programmes Fail: The Realities That Theory Obscures
I have seen well-intentioned AM programmes launched with genuine commitment and then quietly abandoned within twelve to eighteen months. The failure modes are consistent enough to be worth addressing directly.
The most common failure is treating AM as a production department initiative rather than a company programme. AM is owned by Production and supported by Maintenance — that is the right framing. But it must be championed by senior leadership, measured in the same management reviews where output and quality are measured, and resourced adequately in terms of the time operators are given to perform AM tasks during their shifts. When cleaning and inspection time is systematically squeezed out by production pressure, the message that reaches operators is unambiguous: AM is optional. Within months, it is being done cursorily if at all.
The second common failure is neglecting operator training. AM requires operators to develop four distinct equipment-related skills progressively over multiple steps and multiple years. Organisations that launch Step 1 with a cleaning event and never provide the structured technical training of Step 4 end up with operators who are mechanically compliant with a checklist but lack the knowledge to actually detect abnormalities. Cleaning without understanding is janitor work, not autonomous maintenance.
The third failure, particularly common in organisations with a strong throughput culture, is advancing through the seven steps on a management timeline rather than a capability-readiness timeline. Each step has an audit gate — a formal assessment of whether the activities of that step have been fully implemented and the capability genuinely built — before the team progresses to the next. Organisations that rubber-stamp audit gates to stay on schedule end up at Step 5 or 6 on paper while still operating at Step 1 in practice. The audit is not bureaucracy; it provides feedback on strengths and weaknesses, clarifies what needs to be achieved, and determines whether each step has genuinely been accomplished before the foundation is built upon.
The fourth failure is underestimating the cultural dimension. AM is not a technical programme with a cultural component — it is a cultural transformation that happens to use technical tools. The shift from "I operate this machine" to "I own this machine" is a change in identity, and identity changes do not happen because a manager announces a new programme. They happen through sustained practice, genuine recognition, visible management commitment, and the accumulation of small wins that prove to operators that their care for the machine is noticed and valued. Rewards and recognition, success story sharing, and the evolution of the programme over time — starting with a few pilot machines and expanding as capability grows — are not optional add-ons. They are the cultural mechanisms that make the difference between an AM programme that sticks and one that fades.
Measuring What Matters: Metrics and Auditing in AM
A credible AM programme tracks both leading and lagging indicators. The lagging indicators — number of breakdowns, OEE, MTBF, MTTR, maintenance cost, defect rate — tell you whether AM is producing results. The leading indicators — number of tags raised and resolved, number of one-point lessons created, number of skills per operator, number of employee improvement suggestions — tell you whether the behaviours that produce those results are actually happening.
The gap between these two sets of measures is diagnostic. If leading indicators are strong but lagging indicators are not improving, the AM activities are being executed but something in the technical approach or equipment condition is counteracting the gains. If lagging indicators are strong but leading indicators are weak, the results may be fragile — dependent on a few exceptional individuals or on a short-term effort that will not sustain. The healthiest AM programmes show progressive improvement in both.
The JIPM TPM Excellence Criteria for AM provide an external reference for assessing the maturity of an AM programme. Key criteria include: achievement goals being set up and assessed step by step; visual boards showing status of activities, goals, accomplishments, and issues; systematic removal of contamination and its causes; perfect execution of cleaning, lubrication, tightening, and inspection; a functioning skills development system evidenced by skills evaluation charts, one-point lessons, and demonstrated training results; and employee morale surveys and skill maps that confirm genuine engagement. These criteria are not aspirational ideals — they are the standards against which JIPM-recognised plants are actually evaluated, and organisations preparing for TPM excellence awards should use them as a practical implementation checklist from the outset.
The Safety Dimension of Autonomous Maintenance
No discussion of AM is complete without addressing safety explicitly, and the JIPM framework integrates safety into every step of the AM journey rather than treating it as a parallel concern.
Each step of AM has a corresponding safety orientation. Step 1 focuses on identifying and correcting unsafe conditions — exposed moving parts, projecting components, spattering of harmful substances — that initial cleaning reveals. Step 2 addresses problems related to covers and guards. Step 3 establishes and reviews work standards and daily check methods with safety built in. Step 4 includes checking and improving the performance of safety and disposal devices as part of general inspection. Step 5 focuses on correcting stressful working postures and methods that create ergonomic risk. Step 6 ensures workplace organisation maintains a proper working environment. Step 7 cultivates a culture where everyone takes responsibility for the safety of their own workplace.
The connection between safety and AM runs deeper than procedural checklists. Most accidents occur when unsafe conditions and unsafe behaviour coincide. Unsafe conditions — missing guardrails, inadequate safety devices, contaminated surfaces — are precisely the kind of problems that a well-implemented AM programme systematically surfaces and resolves. Unsafe behaviour — actions that result from failure to adhere to specified standards — is precisely what AM's investment in training, visual controls, and daily standards is designed to prevent. AM is therefore not merely complementary to safety management; it is one of its most effective structural supports.
The OEC Autonomous Maintenance Maturity Diagnostic
When I ask plant managers whether they have an AM programme, the answer is almost always yes. When I ask what level they are operating at, the answer is rarely what the project board suggests.
The question worth asking is not "are we doing AM?" but "at what level of maturity are we doing it?" — because the difference between a Level 1 and a Level 4 AM programme is not a matter of degree. It is a matter of whether cleaning is genuinely inspection, whether operators can independently detect and classify abnormalities, whether AM steps have been completed and audited rather than declared, and whether management systems treat AM metrics as leading indicators of equipment health or as compliance checkboxes. The distance between those two states is enormous, and most plants sit much closer to Level 1 than they recognise.
The OEC Autonomous Maintenance Maturity Diagnostic was developed through OEC's consulting practice across semiconductor, automotive, and industrial manufacturing plants in Asia-Pacific. It comprises five dimensions and four levels. Level 1 represents where most plants start — not necessarily because their AM effort is absent, but because the foundational infrastructure of standards, inspection capability, and structured programme management has not yet been built. Level 4 represents the world-class standard that JIPM assessors look for in a TPM Excellence Award submission. The diagnostic is designed to be scored honestly against current reality, not aspiration.
Score each dimension from 1 to 4. Add the five scores together and interpret as follows: a total of 5 to 12 indicates foundational infrastructure gaps — the programme needs structural investment before it can begin to deliver reliable equipment condition improvement. A total of 13 to 16 indicates an intermediate programme with defined gaps — the methodology and intent are present, but specific dimensions need targeted development. A total of 17 to 20 indicates a mature programme — the question at this level is whether AM is genuinely sustaining equipment condition and feeding the FI pipeline, or whether it has become a compliance activity that produces completed checklists without producing equipment health.
Dimension 1 — Basic Equipment Condition and Cleaning Standards
At Level 1, there is no formal cleaning standard for the equipment population. Machines are cleaned when they are visibly dirty or when a scheduled shutdown creates the opportunity, but there is no documented standard specifying what is cleaned, how, with what materials, at what frequency, or by whom. Equipment in this state almost always shows signs of accumulated contamination — oil residue on surfaces that should be dry, debris in recesses that operators have never been asked to clean, deterioration that has progressed unchecked because it has never been made visible. At Level 2, cleaning standards have been drafted and introduced for at least part of the equipment population. Operators are following the standard, but coverage is uneven across shifts and machine types, and the standard itself was written quickly rather than derived from a thorough Step 1 initial cleaning. At Level 3, cleaning standards have been developed through a genuine Step 1 process, cover the full equipment population, specify cleaning methods and materials precisely, and are followed consistently across all shifts. Contamination sources identified in Step 2 have been addressed, and the time required for the daily cleaning routine has been reduced through accessibility improvements. At Level 4, cleaning standards are not merely followed but owned — operators have contributed to refining them, visual controls embedded in the equipment itself make the normal condition immediately observable, and the zero-contamination-source elimination programme has been systematically applied across the equipment population with verified results.
Dimension 2 — Operator Inspection Capability and Abnormality Detection
At Level 1, operators perform their daily tasks but cannot reliably identify abnormalities in the equipment they operate. They know the machine is running or not running; they do not have the technical vocabulary, the baseline understanding of normal equipment conditions, or the structured inspection routine to detect the early warning signs that precede failures. Abnormalities that a trained observer would notice — unusual vibration, elevated surface temperature, a hairline crack in a casing — are invisible to them not because they are incapable but because they have never been given the knowledge or the time to look. At Level 2, inspection awareness is developing. Operators follow a cleaning and inspection checklist and detect obvious abnormalities — visible contamination, audible grinding, leaks — but their detection capability is limited to phenomena that do not require technical understanding to recognise. Many subtle early-warning signals still go undetected. At Level 3, general inspection training under Step 4 has been completed, covering the major equipment subsystems. Operators understand the function of the components they inspect and can interpret their senses diagnostically. The one-point lesson library reflects team-generated knowledge from actual equipment discoveries, and peer-teaching has distributed that knowledge across the team. False-alarm rates — abnormality tags raised for conditions that turn out to be within normal range — are tracked and declining as operators' understanding of normal equipment behaviour deepens. At Level 4, operators autonomously detect, classify, tag, and escalate abnormalities with a reliability and sophistication that approaches that of a skilled maintenance technician for the specific equipment population they operate. False-alarm rates are tracked and trending down, detection frequency — the proportion of incipient failures caught before they become breakdowns — is trending up, and the team's OPL library is actively maintained and used as an onboarding resource for new members.
Dimension 3 — AM Step Progression and Standard-Setting
At Level 1, there is no formal AM step structure. Cleaning and inspection activities may be happening, but they are not referenced to any step framework, they have not been through an audit gate, and there is no mechanism for confirming that the foundational work of one step has been genuinely completed before the next begins. The AM activity board, if it exists, shows the programme as having completed steps that have not in fact been built. At Level 2, the seven-step framework has been introduced and the programme is progressing through it, but step audits are conducted internally without independent verification, and the criteria for step completion are interpreted generously. Steps 1 and 2 may have been declared complete on the basis of a single initial cleaning event rather than a sustained and verified standard. At Level 3, AM steps are progressing through a rigorous audit process — steps are not declared complete until a qualified internal auditor confirms that the required outputs are in place, functioning, and understood by the operating team. The OPL library is maintained with documented dates and the topics generated by the team's own equipment experience. Standards are specific, visual, and usable by an operator with no prior experience of that machine. At Level 4, AM Steps 1 through 7 have been completed and the programme is self-managing. The OPL library is a live knowledge base integrated into new-operator induction, standards are reviewed and updated on a defined cycle and whenever equipment modifications or product changes require it, and the step audit process has been extended to benchmark against comparable plants in the broader organisation or industry.
Dimension 4 — AM-FI-PM Integration
At Level 1, AM operates in isolation. The abnormalities detected and tagged by operators are resolved at the operator or maintenance level, but no mechanism exists to route unresolved or recurring abnormalities into the Focused Improvement pipeline or to update Planned Maintenance schedules based on what AM inspection reveals. The three pillars have their own plans, their own metrics, and their own review cycles; they do not feed each other. At Level 2, informal communication between AM teams and the maintenance department has begun. Some recurring abnormalities that AM teams consistently detect are being noted in maintenance discussions, but the connection is person-dependent and ad hoc rather than structural. At Level 3, a formal abnormality routing process exists: abnormalities that cannot be resolved at the operator level are reviewed by the FI pillar team at a defined frequency, matched against the loss register, and either addressed by planned maintenance or elevated into the FI project pipeline. AM standards are updated when FI countermeasures change equipment configuration or maintenance requirements. At Level 4, AM inspection findings systematically feed the FI project pipeline as structured intelligence about where chronic losses are developing, AM standards are updated to incorporate FI countermeasures as a standard project closure step, and AM inspection routes explicitly incorporate the condition checks that Quality Maintenance analysis has identified as critical to product quality assurance.
Dimension 5 — Management Systems and Governance
At Level 1, there is no AM activity board, no regular management review of AM performance, and no tracking of AM-specific metrics. The programme's existence is asserted in TPM documentation but is not visible in the day-to-day management rhythm of the plant. Production managers can report OEE, breakdown counts, and quality defect rates; they cannot report tag closure rates, abnormality detection frequency, OPL completion rates, or AM step audit scores. At Level 2, an activity board is in place and AM metrics are tracked, but the board is updated infrequently and metrics are not reviewed in management meetings at a frequency that makes them actionable. The programme is visible but not managed. At Level 3, AM performance metrics — tag closure rates, abnormality detection frequency, OPL completion, step audit scores — are reviewed at regular intervals in production and TPM steering committee meetings. Management uses these metrics to identify where additional support or resources are needed and where strong performance should be recognised. AM step audits are scheduled and completed on plan. At Level 4, AM performance metrics are reported as leading indicators of equipment health at the TPM steering committee level, with trend data visible across multiple years. The management cadence treats declining abnormality detection frequency as an early warning of AM programme decay, not as a sign that the equipment is in better condition.
In OEC's experience assessing semiconductor and automotive plants, the modal score across assessed plants is 8 to 10 — which places most programmes solidly in the foundational band, well below the intermediate threshold, and significantly below what JIPM assessors expect in a TPM Excellence Award submission. The diagnostic is not a discouragement. It is a map.
From Compliance to Ownership: What Sustaining AM Really Requires
The question I am asked most often at the end of an AM implementation engagement is some version of: "How do we make sure this doesn't fade away?" The honest answer is that sustaining AM requires the same things that creating it required — active leadership, structured time, genuine capability development, and consistent follow-through — but with the additional challenge of maintaining them when the novelty has passed and the programme faces competition for attention from every other operational priority.
Active leadership for the TPM initiative, with AM owned clearly by the Production department and supported — not directed — by Maintenance, is the non-negotiable starting condition. Proper operator training, certification, and ongoing skills development must be built into the operational rhythm rather than treated as a one-time launch event. The time required for cleaning, inspection, and lubrication must be formally scheduled — not squeezed into gaps or treated as optional — because anything that depends on discretionary time will be the first thing sacrificed under pressure.
Starting with a few pilot machines and building out progressively, sharing success stories visibly, and connecting recognition to AM engagement rather than just output metrics are the cultural practices that make the difference between an AM programme that sustains and one that fades.
The benchmarking data from the 2014 study across ADGT, STATS ChipPAC, and Amkor Technology Philippines illuminated this sustaining gap clearly. Both ADGT and STATS made active use of MTTR and MTBF trend data — not merely as lagging indicators of maintenance performance, but as inputs to skills development programmes and as evidence of where technical gaps required targeted improvement activity. The ability to see that MTBF was declining on a specific equipment population, and to connect that trend back to changes in cleaning consistency, lubrication adherence, or operator skill, is precisely the feedback loop that keeps an AM programme improving rather than merely maintaining. Amkor, by contrast, was not using MTTR and MTBF data at the time of the study — a gap that the report identified explicitly as an area requiring attention. The contrast between the two approaches reflects a broader principle: organisations that treat AM as an activity manage it by tracking whether the activity is happening. Organisations that treat AM as a system manage it by tracking what the activity is producing — in equipment condition, in failure rates, in operator capability, and in the quality of the abnormalities being detected and resolved. The language differed; the organisational discipline required was identical.
And as the programme matures — as operators develop genuine capability, as OEE climbs, as breakdowns become genuinely rare rather than merely less frequent — the AM pillar itself should evolve. The standards become more sophisticated. The inspection capability deepens. The connection to planned maintenance, focused improvement, and quality maintenance strengthens. What begins as a structured programme for building basic operator capability gradually becomes the operating culture of the plant.
That is the destination of Autonomous Maintenance when it is implemented with the seriousness and sustained commitment it deserves: not a programme that runs alongside the work, but a way of working in which every operator is a guardian of the equipment they run, and every machine receives the daily care it needs to deliver its designed performance reliably and safely, day after day, year after year.
About the Author
Allan Ung is the Founder and Principal Consultant of Operational Excellence Consulting, a Singapore-based management training and consulting firm established in 2009. With over 30 years of experience leading operational excellence and quality transformation in manufacturing-intensive environments, Allan's expertise spans Lean Thinking, Total Quality Management (TQM), TPM, TWI, ISO systems, and structured problem solving.
He is a Certified Management Consultant (CMC, Japan), Lean Six Sigma Black Belt, JIPM-certified TPM Instructor (Japan Institute of Plant Maintenance), TWI Master Trainer, ISO 9001 Lead Auditor, and former Singapore Quality Award National Assessor.
During his tenure with Singapore's National Productivity Board (now Enterprise Singapore), Allan pioneered Cost of Quality and Total Quality Process initiatives that enabled companies to reduce quality costs by up to 50 percent. In senior regional and global roles at IBM, Microsoft, and Underwriters Laboratories, he led Lean deployment, quality system strengthening, and cross-border operational transformation.
Allan has facilitated TPM, OEE and Lean programmes for organisations including Temic Automotive (Continental), Analog Devices, Amkor Technology, STATS ChipPAC, Infineon Technologies, Panasonic, Micron, Lam Research, Tokyo Electron, Dorma, and NEC. He holds a Bachelor of Engineering (Mechanical Engineering) from the National University of Singapore and completed advanced consultancy training in Japan as a Colombo Plan scholar.
His philosophy: "Manufacturing excellence is achieved through disciplined systems, capable leadership, and sustained execution on the shopfloor."
His practitioner-led toolkits have been utilized by managers and organizations across Asia, Europe, and North America to build Design Thinking and Lean capability and drive organizational improvement.
For enquiries about Autonomous Maintenance, TPM implementation, or operational excellence consulting, visit www.oeconsulting.com.sg or contact us directly through the OEC website.
Related Articles in the TPM Practitioner Guide Series
Total Productive Maintenance (TPM): A Practitioner's Guide — Hub article covering all eight pillars of TPM, the philosophy behind zero breakdowns, zero defects, and zero accidents, and what it takes to build a sustainable TPM culture.
Planned Maintenance: A Practitioner's Guide — How to design and implement a planned maintenance system that complements Autonomous Maintenance and progressively eliminates unplanned downtime.
Focused Improvement (Kobetsu Kaizen): A Practitioner's Guide — The methodology for targeting, analysing, and eliminating the specific equipment and process losses that hold OEE back.
Quality Maintenance (Hinshitsu Hozen): A Practitioner's Guide — The eight-step methodology for achieving zero defects by establishing and maintaining the precise 4M conditions required to prevent defect generation at the source.
Overall Equipment Effectiveness (OEE): A Practitioner's Guide — A deep dive into the OEE formula, the six big losses, and how to use OEE as a diagnostic tool rather than just a reporting metric.
OEE Benchmarking: A Practitioner's Guide — How to measure, compare, and improve Overall Equipment Effectiveness using structured benchmarking methods drawn from three decades of plant-level experience.
TPM Self-Assessment and the TPM Excellence Award: A Practitioner's Guide — Streamlines the TPM self-assessment and Excellence Award journey, empowering organizations to bridge operational gaps and achieve world-class manufacturing performance.
OEC TPM Maturity Diagnostic: A Practitioner Guide — Bridges implementation gaps with a four-level maturity model based directly on JIPM award checklists, translated into practical descriptors that make the assessment entirely actionable for practitioners.
Build TPM Capability in Your Organisation
At Operational Excellence Consulting, I deliver customised TPM, OEE workshops and implementation programmes for manufacturing organisations across Singapore and the Asia-Pacific region — from foundational two-day workshops to multi-year TPM implementation support, facilitated by a JIPM-certified TPM Instructor.
👉 Explore our TPM training courses and practitioner-led resources:
Operational Excellence Consulting offers a full catalog of facilitation‑ready training presentations and practitioner toolkits covering Lean, Design Thinking, and Operational Excellence. These resources are developed from real workshops and transformation projects, helping leaders and teams embed proven frameworks, strengthen capability, and achieve sustainable improvement.
👉 Explore the full library at: www.oeconsulting.com.sg/training-presentations
© Operational Excellence Consulting. All rights reserved.