Most maturity assessments return a single score for the organisation. A single score cannot be acted on, because predictive maintenance is not funded for an estate: it is funded for transformers, or for pumps. This instrument returns a level for every class of asset you nominate, derived from what demonstrably happens to a measurement rather than from the level you believe you have reached.
Predictive maintenance is not funded for an estate. It is funded for transformers, or for pumps, or for one production line. An organisational maturity score averages those together and returns a number that no budget corresponds to and no individual is accountable for. This instrument returns a level for every class of asset you nominate, derived from what demonstrably happens to a condition measurement rather than from the level you believe you have reached.
The level definitions are quoted from the published Predictive Maintenance 4.0 research by PwC and Mainnovation, June 2017. Level 0 is an Optimal addition, covering the three per cent of the 268 companies surveyed that reported no predictive maintenance at all. The share shown against each level is where those 268 companies sat.
No routine measurement of condition is carried out on this asset class.
What this looks like on site. The class runs to failure, or is maintained on a fixed calendar or runtime interval. Nothing about its current condition feeds the decision to intervene.
Periodic physical inspections; conclusions are based solely on the inspector's expertise.
What this looks like on site. Someone walks the route and looks, listens and touches. A judgement is recorded. No instrument reading is taken, so the conclusion depends entirely on the experience of the person who took it, and is not available to anyone else afterwards.
Periodic inspections; conclusions are based on a combination of the inspector's expertise and instrument read-outs.
What this looks like on site. A technician takes a portable vibration, thermography or oil reading on a route, and interprets the number alongside experience. The reading exists, but only at the moment the route was walked.
Continuous real-time monitoring of assets, with alerts given based on pre-established rules or critical levels.
What this looks like on site. Permanently installed sensors stream continuously, and an alarm is raised automatically when a value crosses a limit that was set in advance. The rule is fixed; the asset is compared against it.
Continuous real-time monitoring of assets, with alerts sent based on predictive techniques such as regression analysis.
What this looks like on site. The same continuous data is used to estimate how long the asset has before it can no longer perform its function, with a stated confidence. The distinction from level 3 is a quantified time to failure rather than a threshold that has been crossed.
A peer comparison is not offered at launch. A benchmark is only as reliable as the cohort behind it, and a cohort of one sector and one region takes time to build. Until it exists, each class is reported against the published ladder and against the 268-company reference distribution above, which is stated rather than implied.
ISO 13374 decomposes condition monitoring into a fixed processing sequence, from taking a reading through to issuing an instruction. Each block is asked once for each asset class you assess. The sequence is what makes the result diagnostic rather than descriptive: it identifies the single block that is constraining a class.
How is the condition of this class actually measured today?
Once a reading is taken, is it retained, aligned to the asset and reachable by the people who need it?
How is a departure from normal identified?
Once a departure is identified, is it diagnosed to a cause, and is that diagnosis held anywhere?
Is the time remaining before functional failure estimated, and on what basis?
Does the finding become maintenance work, with the parts and the urgency attached to it?
Advisory Generation is deliberately excluded from the level. An asset class can measure condition well and convert none of it into maintenance work. Including it within a single level would conceal it. It is reported separately instead, and where a class measures condition but routes nothing, the response is flagged as Stranded Measurement — usually the least expensive finding in the report to resolve, because the measurement has already been paid for.
The rules are evaluated from the top down, and the first level whose conditions are met is the level reported. Nothing is averaged, and no level is accepted because it was claimed. The mapping from the six blocks to the ladder is Optimal's own inference; it is not published by PwC or Mainnovation, and every report says so on its methodology page.
| Level returned | What must be evidenced for that level to be reached |
|---|---|
| Level 4 | Continuous installed acquisition, data aligned and reachable, automated state detection, with both diagnosis and prognosis carried out in some form |
| Level 3 | Continuous installed acquisition, data aligned and reachable and automated state detection against rules or critical levels |
| Level 2 | At least portable instrument readings, data retained in some form and some comparison against a baseline or limit |
| Level 1 | Some routine measurement of condition, at minimum by human senses |
| Level 0 | No routine measurement of condition on this class |
A positive gap is not an error and is not treated as one. It is the measurable distance between how an estate describes itself and what the evidence supports — recorded per asset class, which is the level at which it can be discussed and closed.
Twelve questions are asked once for the organisation. They set the level any class can reach and sustain, whatever the instrumentation fitted to an individual class. A class may well be developed above this ceiling. When it is, the capability rests on particular individuals rather than on established practice, and it is not reliably retained when those individuals leave. The report records that as a continuity risk rather than as an error.
Whether the asset population is captured, and whether a hierarchy and naming standard is genuinely followed rather than merely published.
Whether failure modes are defined for the critical classes, and how much of that population has been through FMEA or RCM.
The depth and usability of work order history, and how consistently failures are coded and against what.
The coverage, retention and accessibility of the process historian and the wider condition data estate.
How a monitoring finding actually reaches the work management system, and whether that path is automatic or manual re-entry.
The analytical capability available to interpret what is collected, and whether it is retained in the organisation.
Each layer is scored out of five and read as the level it supports. The lowest layers set the ceiling. A question answered not known is excluded from its layer average rather than scored as zero, so no organisation is penalised for a gap in knowledge.
Recommendations are ordered rather than listed. Each is scored on what it would enable, what it would cost to carry out and how confident the underlying answers were. The weightings are Optimal's own and are printed in the method appendix of every report, so the order can be challenged.
A block answered not known is excluded from the derivation rather than scored as zero. A class is never recorded at a lower level because a respondent declined to estimate.
Flags are raised by the scoring service wherever the pattern of answers is internally inconsistent, or where a known failure mode in predictive maintenance programmes is present. They are observations presented for review, not conclusions drawn about the organisation.
Claimed Capability Not Supported
The class was placed higher than the evidence derives.
Stranded Measurement
Condition is measured at level 2 or above, and nothing is routed to a work order.
Sensing Without Interpretation
Sensors are installed and streaming, but nothing compares the output against a baseline.
Prognosis Without Diagnosis
Time to failure is being estimated for a class whose faults are not diagnosed.
Data Not Reachable
Readings are being taken and are not retained in any usable form.
Capability Ahead of Organisation
A class sits at level 3 or above while the organisational foundation beneath it does not support that.
Organisation Ahead of Capability
The organisation is well run and disciplined, and no asset class has been taken beyond level 1.
The result is not a scorecard with a covering note. Every asset class you assess receives its own chapter, with the block-by-block assessment, the constraining block named, the full answer record and what would move it. The result is complimentary and carries no obligation.
Executive Summary
Result overview, respondent and scope
Asset Class Scorecard
Every class assessed, side by side
Claim Gap Breakdown
Where the estate describes itself differently from the evidence
Standards Alignment
What the method is based on, and what is Optimal's own
Foundation
The organisational basis for every class
Asset Class Analysis
Block-by-block assessment, a chapter for each class
Prioritised Recommendations
What to do first, and why that order
Risk Flags
Patterns raised automatically for review
Ladder Position
Where each class sits against the published ladder
Longitudinal Trends
What the second assessment cycle will compare
Advancement Pathway
A roadmap sequenced into your planning horizons
Propensity to Act
Constraint, appetite, horizon and purpose
Methodology
Derivation, weighting and the treatment of missing data
Every report carries this table. Where the method is based on a published standard, the standard is named. Where it is based on Optimal's own inference, that is stated as plainly as the rest, because a method that cannot be separated into those two parts cannot be properly examined.
| Source | What it supplies | Status |
|---|---|---|
| PdM 4.0 ladder | The five levels and their definitions, from PwC and Mainnovation, June 2017 | Published |
| ISO 13374 | The six functional blocks and the order they occur in | Published |
| ISO 13381-1 | Prognostics: what a time to functional failure means, and how confidence is expressed | Published |
| ISO 17359 | Condition monitoring and diagnostics of machines, general guidelines | Published |
| ISO 14224 | Equipment-class naming used for the asset classes | Published |
| ISA-95 / IEC 62264 | Fixes the level a response is given at, so two answers mean the same thing | Published |
| Block-to-level derivation | The mapping from the six blocks to a ladder level | Optimal's own inference |
| Foundation ceiling | The maximum level the organisational foundation is judged to sustain | Optimal's own inference |
| Priority weighting | Consequence, magnitude, confidence and effort weights used to sequence the roadmap | Optimal's own inference |
The two instruments answer different questions and are designed to be read together. Where a GARPI composite already exists it is read as context here and never recalculated. Where a class derives well above what the organisation around it can sustain, that is what the two instruments together are intended to identify.
The GARPI™ BenchmarkProgress is saved against a verified work email, so you can stop at any point and return to exactly where you left off, or pass the remainder to a colleague who knows a class better than you do.
Verify a work email
A single link confirms who is answering and secures your progress. Nothing is published against you or your organisation.
Answer twelve foundation questions
Asked once for the organisation. They establish the ceiling that applies to every class you go on to assess.
Nominate the classes that matter
Filtered to your sector, named to ISO 14224. There is no limit on how many you assess, and thirteen questions follow for each one.
Receive a reviewed report
Scored by the server rather than in your browser, then reviewed against the method before it is issued to you.
Running this in a workshop. A facilitation mode records who supplied each answer and captures a note against any question where the discussion went further than the options allowed. Those notes are carried into the report rather than discarded, which is usually where the most useful material in a workshop is produced.
There is no limit on how many asset classes you assess, and you can stop at any point and return later.