Why WCM
A methodology is a loop. Most software cuts it in the middle.
World Class Manufacturing and TPM are not reporting frameworks. They are a set of loops that only pay off if they close — and closing them is a software problem long before it is a discipline problem.
The short version
What WCM and TPM actually ask of a plant.
Ten pillars, a scored audit behind each one, and a cost lens over the whole thing.
Total Productive Maintenance made a claim that sounded strange in 1971 and is now unremarkable: equipment reliability is not the maintenance department’s private problem. Operators own the basic conditions of their machines. Maintenance owns the deterioration they cannot prevent. Engineering owns the weaknesses that keep coming back.
World Class Manufacturing took that idea, added a cost lens, and organised the whole thing into ten pillars with a scored audit behind each one. The mechanism underneath is unglamorous. Every loss is classified and given a monetary value. Every valued loss is attributed to exactly one pillar. Each pillar climbs a ladder of seven steps, and it climbs by producing evidence rather than by declaring progress.
A record is not finished when the machine runs again. It is finished when a standard has changed.
That last sentence is the whole thing. A breakdown that produced a repair produced nothing. A breakdown that produced a shorter lubrication interval, a revised standard maintenance procedure, a one-point lesson for the operators on the line and a cost entry against the right pillar has produced everything — and those four artefacts are what the assessor asks to see.
- EWO
- Emergency Work Order — the structured breakdown record: phenomenon, six MTTR phases, 5W1H, 5-why, root cause, countermeasures.
- Machine ledger
- The per-component 52-week maintenance calendar that plans the year and records what actually happened.
- Cost deployment
- The A–F matrix chain that turns losses into money, money into projects, and projects into verified savings.
- PQCDSM
- The criticality rubric — production, quality, cost, delivery, safety, morale — that decides which machines get which regime.
- CD-year
- The cost-deployment year. Baselines, matrices, boards and checklists are frozen against it so an audit can be replayed.
- Blue room
- The review room where pillar boards are read. Historically a week of PowerPoint per pillar leader.
The Professional Maintenance ladder
Seven steps, climbed with evidence.
A plant does not have “a PM step”. Every machine has one, with planned and actual dates and an audit checklist that decides whether the claim stands.
- 01Eliminate forced deterioration
- Restore basic conditions first — cleaning, lubrication, tightening — and stop the damage the plant is doing to itself. Nothing measured before this step means very much.
- 02Reverse deterioration
- Analyse every breakdown rather than the memorable ones. Classify the causes, stratify the repair, and measure MTBF and MTTR from records instead of from recollection.
- 03Set the maintenance standard
- Write down what maintaining this component actually means: the procedure, the parts, the time, the skill. This is where the machine ledger and the SMP library come from.
- 04Counter the weak points
- Where the failure is a design weakness, the countermeasure is not a shorter interval — it is feedback to engineering. Component life gets extended, and the next machine inherits the lesson.
- 05Build the time-based system
- Intervals derived from the failure series rather than copied from the vendor manual. This is where MTBF − 2σ and σ/MTBF stop being formulas and start being decisions.
- 06Build the condition-based system
- Migrate the lines where a reading predicts better than a calendar — behind a cost/benefit gate, because a sensor that saves less than it costs is a hobby.
- 07Manage the cost of maintenance
- Overmaintenance becomes measurable — what the plan costs at the current step versus at step one — and the programme optimises toward zero breakdowns rather than toward more inspections.
This is why a step claim is a place software can be genuinely useful. A machine cannot credibly be at step five while its trailing six months show four repeat breakdowns on the same component — and that is a rule, checkable, with the failure history right there.
The learning-loss argument
A generic CMMS does not break the method. It quietly stops recording it.
Free text has no shape, and a loop cannot close through a field nobody can query.
In a generic CMMS
Work order closed. Notes: “bearing seized, replaced, line back up 14:32.” Accurate, useful to the person who wrote it, and unqueryable by anything.
In an EWO
Phenomenon, stoppage window, MTTR split across six phases, 5W1H, five parallel 5-why chains terminating at a typed root cause, and countermeasures whose own work has to close first.
Six months later
The same bearing fails again. Nothing links the two events, so the interval is never questioned and the second failure is filed as bad luck.
Six months later
The repeat is visible on the component’s own row, the trailing failure series has moved, and the suggested interval moves with it.
At the year-end review
Downtime cost is reconstructed in a spreadsheet from exports, and the number depends on who built the sheet.
At the year-end review
The loss was valorised when it happened, against the pillar that owns it and the node it happened on. The matrix is a view of records.
What was lost
Not the repair — the learning. The plant is exactly as good at preventing that failure as it was before it happened.
What was gained
A shorter interval, a revised procedure, a trained operator and a cost entry — produced as a side effect of closing the record.
This is also why AI on top of a generic CMMS hits a ceiling. A copilot over free-text work orders can summarise and suggest; it cannot run a σ/MTBF gate or verify a 5-why against live failure history, because the data it would need was never captured in a shape it could read. Ours does both, and does the second one by fetching the actual chain, pseudonymising it, and returning sound, weak or restates the symptom with the exact words it was shown. The agent works because the model does.
Audit readiness
The evidence pack should already exist when the assessor arrives.
None of the collecting creates value. All of it exists somewhere already.
Ask any pillar leader what the weeks before an assessment look like. They are spent collecting: screen captures, before-and-after photographs, signed checklists, training records, calibration certificates, the KPI trend with the awkward month explained.
Treating audit readiness as a feature changes what the software has to do. Evidence is collected as work happens and attached to the step that mandates it. Checklists and scoring rubrics are versioned and snapshotted per CD-year, so a scored audit can be replayed exactly as it was scored rather than as today’s configuration would score it.
And once the pack exists continuously, it can be read early. A dry-run before the real assessment produces a findings report with a predicted pillar score. The unpleasant surprises move from the assessment to a Tuesday, which is where they are cheap.
The same argument covers the IATF 16949 and ISO 50001 obligations that sit next to the WCM programme: competency evidence, calibration and gauge genealogy, energy baselines. They are not compliance modules bolted on. They are the ordinary output of a system that records who did what, with which instrument, against which standard.
What the dry-run looks for
- Top losses with no improvement project attached
- KPI actuals that have gone stale against the route map
- Route-map targets missing for the audit year
- Evidence packs missing a mandatory artefact type
- Step claims that the failure history does not support
- EWOs closed without a validated root cause
Scored audit runs, the step-claim register, the controlled-document register and the agent checks behind them all run today, and an audit run exports as a scored workbook — weighted in integer tenths, where one critical NOK fails the audit regardless of the percentage.
If this is your world
You already know whether this argument is true.
If you run a pillar, none of the above needed explaining — which means the useful conversation is about your ledger, your EWO backlog and your next assessment date, not about what WCM is.