A predictive maintenance program that cannot measure itself cannot improve itself. Yet a significant number of manufacturing facilities that have invested in PdM platforms — sensors, AI models, and mobile dispatch workflows — continue to track only the metrics they inherited from their legacy preventive maintenance culture: scheduled PM compliance and breakdown work order count. These are necessary but insufficient. A mature PdM program demands a distinct KPI architecture that measures prediction quality, program health, asset reliability, and financial return simultaneously. Without that structure, reliability teams have no objective basis for proving program value to leadership, identifying where the AI model is underperforming, or knowing when the program is ready to scale beyond the pilot assets. iFactory AI provides a live KPI dashboard that tracks all 12 critical PdM performance metrics in real time — from MTBF and MTTR trending by asset class to false-positive rate and PdM work order ratio — giving plant leadership and reliability engineers the same data layer in one platform. Book a Demo to see how the dashboard maps against your current measurement framework.
Track Every KPI That Proves Your PdM Program Is Delivering
iFactory AI provides a unified KPI dashboard covering MTBF, MTTR, OEE, prediction accuracy, false-positive rate, and maintenance cost per unit — updated in real time from live asset and work order data.
The 4-Layer PdM KPI Framework: What to Measure and Why
The 12 KPIs that define a world-class PdM program fall into four distinct measurement layers, each answering a different question about program health. Lagging reliability indicators like MTBF and OEE tell you what already happened to your assets. Leading program indicators like PdM work order ratio and scheduled compliance predict whether the program will sustain itself. Prediction quality metrics — false-positive rate, detection rate, and prediction lead time — evaluate the AI model itself. And financial metrics close the loop by connecting maintenance performance to the P&L. Reliability managers building or auditing their KPI architecture frequently Book a Demo with iFactory to align their current dashboard against this four-layer structure and identify measurement gaps.
Asset Reliability
Lagging indicators measuring actual equipment performance outcomes. These tell you whether your assets are becoming more reliable over time.
Program Health
Leading indicators measuring how proactively and consistently the program is being executed. These predict future reliability performance.
Prediction Quality
Metrics that evaluate the AI model directly. If prediction accuracy is high but ROI isn't improving, the bottleneck is execution, not the technology.
Financial Return
Metrics that translate maintenance outcomes into the business language that finance and plant leadership use to evaluate program investment.
Defining and Benchmarking Each PdM KPI
The table below defines each of the 12 essential PdM KPIs, provides the calculation formula, and maps it against world-class performance benchmarks drawn from SMRP Best Practices and industry data for U.S. discrete manufacturing operations. iFactory's dashboard tracks and trends all 12 automatically from work order and sensor data — no manual spreadsheet calculation required.
| KPI | Layer | Formula / Definition | Typical Plant | World Class (SMRP) | iFactory Tracking |
|---|---|---|---|---|---|
| Mean Time Between Failures (MTBF) | Reliability | Total uptime ÷ number of failures in period | Varies widely by asset class | Trending upward, quarter over quarter | Tracked per asset, line, and site in real time |
| Mean Time To Repair (MTTR) | Reliability | Total repair time ÷ number of repair events | 2–8 hours (manufacturing average) | <1.5 hrs for high-criticality assets | Auto-calculated from work order open/close timestamps |
| Asset Availability | Reliability | MTBF ÷ (MTBF + MTTR) × 100 | 78–85% | ≥95% for critical assets | Live availability trending per asset and plant zone |
| Overall Equipment Effectiveness (OEE) | Reliability | Availability × Performance × Quality | 60–65% | ≥85% (world class benchmark) | Availability component auto-fed from downtime work orders |
| PdM-to-Total Work Order Ratio | Program | PdM work orders ÷ total work orders × 100 | <15% (most plants) | ≥30% indicates mature PdM program | Work order type auto-tagged; ratio reported on KPI dashboard |
| Planned Maintenance Percentage (PMP) | Program | Planned maintenance hours ÷ total maintenance hours × 100 | 55–65% | ≥85% (SMRP world-class threshold) | All work order types classified at creation; PMP calculated automatically |
| Schedule Compliance Rate | Program | Work orders completed on schedule ÷ total scheduled × 100 | 70–80% | ≥90% for critical A-class assets | Schedule vs. actual completion tracked per work order in EAM |
| Detection Rate (True Positive Rate) | Prediction | Failures predicted in advance ÷ total actual failures × 100 | 50–65% (rule-based systems) | ≥85% (AI-driven systems) | Confirmed catch events logged and trended against total failures |
| False Positive Rate | Prediction | False alarms ÷ total alerts generated × 100 | 25–40% (early deployments) | <10% (mature AI model with baseline validation) | Debounce logic and sustained-threshold rules minimise FPR from deployment |
| Prediction Lead Time | Prediction | Average time between alert and actual failure event | Hours to days (reactive alerts) | ≥14 days for parts procurement and scheduling | Lead time measured per alert event and trended by asset class |
| Maintenance Cost Per Unit of Output | Financial | Total maintenance spend ÷ units produced in period | Varies by industry; tracks against baseline | Year-over-year downward trend after PdM scale-up | Maintenance cost data from EAM synced with production volume data |
| Downtime Cost Avoided (PdM ROI) | Financial | (Predicted failures × avg. downtime cost) − PdM program cost | Positive within 6–18 months for most facilities | 10:1 to 30:1 ROI within 12–18 months | ROI model built from work order data; reported in plant leadership dashboard |
The KPIs Most Programs Measure Wrong — and How iFactory Fixes Them
Three of the 12 KPIs are consistently misunderstood or miscalculated in manual tracking environments: MTTR, PdM-to-total work order ratio, and false-positive rate. Each has a common measurement error that causes programs to appear more mature than they actually are — and each is automatically calculated correctly inside iFactory's EAM platform. Reliability teams who Book a Demo regularly find that their current self-reported KPIs shift materially once automated calculation replaces manual spreadsheet tallying.
Most teams measure MTTR from the moment a technician starts the repair to when the machine restarts — excluding detection time and parts procurement delays. This systematically understates true repair burden and masks workflow inefficiencies upstream of the wrench-on-bolt moment.
iFactory captures MTTR from work order creation timestamp (the moment the alert triggers dispatch) to close timestamp (scan-to-close on mobile). This includes detection-to-dispatch latency, parts pull time, and active repair time — giving a complete picture of total response burden.
Plants count any work order that references a sensor alert as a "PdM work order" regardless of whether the alert was condition-triggered or simply a time-based PM inspection that happened to use a sensor. This inflates the PdM ratio and overstates program maturity.
iFactory's work order engine classifies work orders at origin: only orders auto-generated by a condition threshold breach qualify as PdM. Calendar-scheduled inspections, even on sensor-equipped assets, are classified as preventive — keeping the PdM ratio accurate and defensible.
Teams that don't formally log technician findings on alert-triggered work orders have no way to distinguish true catches from false alarms. Without this data, false-positive rate is unmeasurable — and chronic alarm fatigue builds silently until technicians stop responding to alerts.
Every iFactory work order requires a findings field at close. The technician logs whether the predicted condition was confirmed, partially confirmed, or not found. This closed-loop feedback automatically calculates false-positive rate by asset class and continuously retrains threshold logic to reduce it.
Where World-Class PdM Programs Stand: Industry Benchmark Comparison
The bar chart below maps typical U.S. manufacturing plant performance against SMRP world-class benchmarks across the four most commonly reported KPIs. The gap between "typical" and "world class" is not a technology gap — it is a measurement and program execution gap. iFactory closes it by automating the data collection, calculation, and trending that most plants are currently doing manually or not at all.
"The question we always ask when a new plant comes to us is: which of the 12 KPIs are you actually measuring, and how are you calculating them? In almost every case, the answer is that they're tracking MTBF on a whiteboard, OEE in a spreadsheet, and nothing in the prediction quality layer at all. False-positive rate and prediction lead time are the two most diagnostic KPIs for a PdM program, and they're almost universally absent. When we bring those into the iFactory dashboard alongside the reliability and financial metrics, the team gets a complete picture for the first time — and the improvement priorities become obvious immediately."
From Data Collection to Program Intelligence: How iFactory Closes the KPI Loop
A predictive maintenance program measured only by reactive outcomes — breakdown count and repair cost — is not a PdM program. It is a reactive program with sensors attached. The 12 KPIs mapped in this guide form a complete measurement architecture: reliability indicators confirm that assets are improving, program indicators confirm that the PdM workflow is executing consistently, prediction quality metrics confirm that the AI model is trustworthy, and financial metrics confirm that the investment is returning value. iFactory AI automates the calculation of all 12 from live work order data and sensor streams — removing the manual spreadsheet burden that prevents most reliability teams from tracking the metrics they know matter. The result is a plant leadership dashboard and a reliability engineer dashboard that tell the same story with the same numbers, updated in real time. Book a Demo and walk through how your current KPI data maps into the iFactory measurement framework.
PdM KPIs and Program Measurement — Frequently Asked Questions
What is a world-class MTTR target for manufacturing?
Per SMRP benchmarks, world-class MTTR for high-criticality assets targets under 1.5 hours from alert to close; iFactory captures this automatically from work order timestamps including detection-to-dispatch latency, not just active wrench time, giving a complete and defensible MTTR figure.
What false-positive rate should a mature PdM AI model achieve?
A mature AI-driven PdM model with a validated baseline should achieve a false-positive rate below 10%; early deployments without a baselining window commonly see 25–40%, which is why iFactory mandates a sustained-threshold baselining period before activating automated work order dispatch.
How does iFactory track PdM-to-total work order ratio?
iFactory classifies work orders at origin — only orders generated by a real-time condition threshold breach qualify as PdM; calendar-scheduled PMs on sensor-equipped assets are correctly classified as preventive, keeping the ratio accurate and preventing the metric inflation common in manual tracking environments.
Which KPI should a program prioritize first when starting out?
Planned Maintenance Percentage is the highest-leverage early KPI — increasing PMP from a typical 55–65% toward the 85% world-class threshold produces the fastest MTBF improvement and provides plant leadership with a clear, directional metric that demonstrates program progress before financial ROI is fully visible.
Can iFactory's KPI dashboard integrate with SAP or existing CMMS data?
Yes — iFactory features pre-built bidirectional connectors for SAP PM, Oracle EAM, and major CMMS platforms, allowing historical work order data to seed baseline KPI calculations immediately at deployment rather than waiting 6–12 months to accumulate sufficient data for trend analysis.
Stop Estimating Program Value. Start Measuring It in Real Time.
iFactory AI automates all 12 PdM KPIs from live work order and sensor data — giving reliability engineers and plant leadership the same measurement layer, updated continuously, without a single spreadsheet.
