Advanced Manufacturing KPIs: 2026 Field Guide (OEE, FPY, Beyond)

Last Updated: 2026-05-18

Every factory I have ever walked into has a wall of KPIs. Usually too many of them, usually contradictory, and usually one or two that nobody can quite agree on the formula for. Advanced manufacturing KPIs are the small set of metrics that actually drive smart-factory decisions in 2026 — overall equipment effectiveness, total effective equipment performance, first-pass yield, rolled throughput yield, mean time between failures, mean time to repair, energy intensity per good unit, and a handful of people-and-safety measures. This field guide is the working definition I use with operations leaders when we sit down to tear up the dashboard and rebuild it.

In 60 seconds — what are advanced manufacturing KPIs? Advanced manufacturing KPIs are quantitative measures that capture output, quality, maintenance, energy, and people performance across a production system in a way that supports decisions, not just reporting. They are grounded in standards such as ISO 22400, VDI 5600, and ISA-95, normalised per good unit produced, and computed in near real time from MES, historian, and CMMS sources. Done right, they collapse hundreds of raw signals into a dashboard a plant manager can act on in one shift.

Most factories still over-index on OEE and treat it as a single number to chase. The five-bucket framework below is the alternative: a small, balanced set of KPIs that does not let one dimension hide a problem in another.

The KPI Framework: Output, Quality, Maintenance, Energy, People

A useful manufacturing KPI framework groups metrics into five buckets — Output, Quality, Maintenance, Energy, and People — each answering a different operational question. Output measures how much the system produced against capacity. Quality measures how much of that output customers will accept. Maintenance measures whether the equipment can be relied on. Energy measures the cost and emissions per unit. People measures the workforce that keeps the system running. Any KPI not landing in one of those buckets is probably a vanity metric.

Why five buckets and not three or seven? Three is the legacy view — Availability, Performance, Quality, which is just the inside of OEE. It misses entire categories of risk. Seven invites duplication and dashboard sprawl. Five maps onto the way operations leaders actually allocate weekly attention: one stand-up on output, one walk-down on quality, one PM review with maintenance, one energy and emissions check, and one safety/people huddle. If you can run a Monday-morning meeting with one slide per bucket, the framework holds.

The other reason for five is alignment with ISO 22400, the international standard for KPIs in manufacturing operations management. ISO 22400-2 enumerates thirty-four key performance indicators across production, throughput, maintenance, and quality, and a separate set for energy. The five-bucket framework collapses those thirty-four into a navigable taxonomy: Output covers production volume, throughput, and capacity utilisation; Quality covers FPY, RTY, scrap, and DPMO; Maintenance covers MTBF, MTTR, MDT, and PM compliance; Energy covers kWh/unit and carbon intensity; People covers TRIR, training hours per FTE, and absenteeism. Each bucket has at most five canonical KPIs in any dashboard.

A quick reference table for the canonical KPIs in each bucket:

Bucket	Canonical KPIs	Primary source	ISO 22400 reference
Output	OEE, TEEP, Throughput, Cycle Time, Capacity Utilisation	MES, Historian	5.4.1, 5.4.2, 5.5.1
Quality	FPY, RTY, Scrap Rate, Rework Rate, DPMO	MES, QMS	5.6.1 — 5.6.4
Maintenance	MTBF, MTTR, MDT, PM Compliance, Availability-A	CMMS, Historian	5.7.1 — 5.7.4
Energy	kWh per good unit, kg CO2e per good unit, Water per unit, Compressed-air leakage	EMS, sub-meters	ISO 50006, ISO 22400 extension
People	TRIR, training hours/FTE, absenteeism, suggestion-rate	HRIS, EHS	Not in 22400, supplementary

Pick three to five KPIs per bucket and resist the urge to add more. The most common dashboard failure I see is forty KPIs spread across three monitors and no one looking at any of them.

OEE and TEEP: Definitions, Formulas, Pitfalls

OEE answers a single question — of the time you planned to run this asset, what percentage produced sellable output? It is the most-used and most-misused advanced manufacturing KPI in 2026, partly because the formula is deceptively simple and partly because every vendor calculates it slightly differently.

The canonical OEE formula is the product of three rates:

OEE = Availability x Performance x Quality

Where:

• Availability = Run Time / Planned Production Time
• Performance = (Ideal Cycle Time x Total Count) / Run Time
• Quality = Good Count / Total Count

The waterfall view in Figure 1 makes the arithmetic concrete. Start with planned production time — typically a shift minus breaks and planned maintenance. Subtract availability losses (breakdowns, unplanned changeovers, material starvation) to get run time. Subtract performance losses (minor stops under six minutes, reduced speed running) to get net run time. Subtract quality losses (scrap units and time spent producing rework) to get fully productive time. Divide fully productive time by planned production time and you have OEE.

Figure 1: OEE waterfall — Availability x Performance x Quality = 74.8% on a typical shift.

World-class benchmarks are widely quoted at 85%, originating from Seiichi Nakajima’s TPM work in the 1980s and propagated through Vorne, iSixSigma, and the VDI 5600 handbook. The 85% benchmark decomposes as 90% availability x 95% performance x 99.9% quality. Most discrete manufacturers run 50-65% OEE; process industries with continuous flows run higher, often 75-85%. Treat the 85% benchmark as a discrete-manufacturing aspiration, not a universal target.

TEEP — Total Effective Equipment Performance — extends OEE by including unplanned shifts as available time. Where OEE uses planned production time as the denominator, TEEP uses all 168 hours in a week (TEEP = OEE x Utilisation, where utilisation is planned production time divided by total calendar time). A line running OEE 80% on one shift per day has TEEP of roughly 27%. TEEP is the metric you reach for when sales ask “can we double production without buying another line?” The answer is in the gap between current TEEP and 100%.

The common OEE pitfalls in 2026 are predictable. First, denominator drift. Teams quietly remove changeovers, breaks, or shutdown time from planned production time to inflate availability. A shift’s OEE goes from 65% to 80% with no actual improvement. Defend the denominator. Planned production time should be the time the line is scheduled to make sellable product, no more and no less.

Second, ideal cycle time inflation. Performance compares actual cycle time to an ideal. If the ideal is set generously, performance looks good even when the line is slow. The ideal cycle time should come from machine OEM specifications or from the best sustained ten-minute run observed during commissioning — not from the average of the last quarter.

Third, treating OEE as a single number. OEE 75% can mean Availability 95% x Performance 79% x Quality 100%, or it can mean Availability 79% x Performance 95% x Quality 100%. Same headline, completely different problems. Always report the three components alongside the rollup, and dashboard them as a stacked time series so the pattern of losses is visible.

Fourth, ignoring micro-stops. Stops under six minutes are usually charged against performance, not availability — but most data-acquisition systems either miss them entirely or aggregate them incorrectly. A properly wired unified namespace architecture with per-asset state tagging makes micro-stops visible. Without it, performance loss masquerades as “ideal-but-slow” and the root cause never gets attention.

A correctly computed OEE, supported by waterfall visualisation of the three components, is one of the strongest signals in operations. A wrong OEE — denominator-inflated, ideal-cycle generous, micro-stops missing — is worse than no OEE because teams act on it as if it were real.

First-Pass Yield (FPY) and Rolled Throughput Yield (RTY)

OEE captures equipment effectiveness but it bundles quality into a single percentage. For multi-stage production lines, quality needs a finer instrument. FPY and RTY are that instrument.

First-pass yield (FPY) is the proportion of units that pass a stage on the first attempt, with no rework. Not the proportion that are eventually good after rework — the proportion that are good without rework. The formula is:

FPY = Units exiting a stage as good first time / Units entering the stage

Many shop floors compute “yield” as final-good divided by total-input, which conflates first-pass and reworked output. That conflation hides the rework loop — the “hidden factory” Joseph Juran wrote about in the 1950s — that consumes labour, tooling, energy, and floor space without showing up on a single KPI.

Rolled throughput yield (RTY) is the product of FPY across all stages of a multi-stage process. If a line has five stages with FPYs of 98%, 96%, 94%, 97%, and 99%, the RTY is:

RTY = 0.98 x 0.96 x 0.94 x 0.97 x 0.99 = 0.849

A line where every stage runs above 94% FPY still has only an 85% RTY because the small per-stage losses compound. RTY is brutal in this way — and that brutality is what makes it useful. It is the KPI that captures the cumulative impact of quality losses end-to-end. Figure 2 shows the propagation.

Figure 2: FPY at each stage compounds into an end-to-end RTY of 84.9% with a visible rework loop.

The relationship to defects per million opportunities (DPMO) and Six Sigma levels. RTY can be expressed as a sigma level. An RTY of 99.9997% corresponds to roughly six sigma — 3.4 defects per million. The 84.9% RTY above is around 2.5 sigma, which is a fairly typical state for a non-Six-Sigma operation. Tracking the sigma equivalent of RTY in addition to the percentage gives leaders an intuitive scale for how far improvement needs to go.

Why RTY beats final-yield reporting. A plant that quotes “98% yield” because final acceptance is 98% may have an RTY of 70% — meaning 30% of units required rework that quietly added to cycle time, labour cost, and material loss. That 28-point gap is the hidden factory. Switching the dashboard from “final yield” to “first-pass yield + rolled throughput yield” makes the hidden factory visible and gives quality engineers a concrete target.

Sampling and measurement system analysis matter. FPY and RTY are only as good as the inspection system that classifies units as good or bad. A high-noise gauge calls good units bad and vice versa, biasing FPY in unpredictable ways. Before trusting an FPY number, confirm the gauge R&R study is current and that the inspection frequency is sufficient to detect short-term excursions. ISO 22400-2 includes data-quality assertions for exactly this reason.

Pair RTY with scrap and rework cost. A KPI in percentage terms is harder to defend in a budget meeting than the same KPI in currency. Translate RTY losses into monthly scrap-and-rework cost: Cost = (1 - RTY) x total units x average unit margin. A 15% RTY gap on a line producing 100,000 units a month at $20 margin each is $300,000 a month of recoverable value. CFOs respond to that number.

FPY and RTY together turn quality from a single number on a dashboard into a stage-by-stage map of where defects originate, how they propagate, and what they cost. They belong in every advanced manufacturing KPI suite.

Maintenance KPIs: MTBF, MTTR, and OEE-A vs MTBF Coupling

Maintenance KPIs feed directly into OEE availability, and they tell a story OEE alone cannot tell. Mean time between failures captures asset reliability; mean time to repair captures organisational maintainability. The two are coupled and need to be read together.

MTBF — mean time between failures — is the average operating time between two consecutive failures. Formula: MTBF = Total uptime / Number of failures. If a machine ran 720 hours in a month and failed three times, MTBF is 240 hours. MTBF is a reliability metric; higher is better; it improves with better components, condition-based maintenance, and predictive analytics.

MTTR — mean time to repair — is the average time from failure detection to restored operation. Formula: MTTR = Total downtime / Number of failures. If those three failures consumed nine hours of downtime in total, MTTR is three hours. MTTR is a maintainability metric; lower is better; it improves with spare-parts availability, diagnostic skills, and modular design.

Availability from MTBF and MTTR:

Availability = MTBF / (MTBF + MTTR)

This is the same Availability that feeds OEE. A line with MTBF of 240 hours and MTTR of 3 hours has availability of 240 / 243 = 98.8%. Lift MTBF to 500 hours with no change in MTTR and availability rises to 99.4%. Cut MTTR to one hour with no change in MTBF and availability rises to 99.6%. Improvement of either lever drives OEE-A directly, and Figure 3 shows the cycle.

Figure 3: The MTBF / MTTR cycle and the availability formula that links them to OEE.

MDT — mean downtime — extends MTTR. Where MTTR measures the time the technician is hands-on, MDT measures total downtime including detection delay, dispatch lag, parts wait, and verification. A factory with 30-minute hands-on repairs but two-hour dispatch lags has poor MDT despite good MTTR. MDT exposes organisational friction that MTTR alone hides.

PM compliance percentage. What proportion of scheduled preventive maintenance tasks were completed on time during the period? PM compliance below 90% is a leading indicator that MTBF will fall, because deferred PMs accumulate latent failures. Track PM compliance as a leading indicator and MTBF as a lagging one.

The OEE-A vs MTBF coupling pitfall. A common trap: a plant reports OEE-A of 95% and MTBF of 40 hours and treats both as healthy. They are inconsistent. 95% availability over a 40-hour MTBF implies MTTR of two hours, which is plausible — but if MTTR is actually 10 minutes, then OEE-A and MTBF cannot both be true on the same data. The discrepancy is usually because OEE-A excludes some downtime category (changeovers, planned maintenance) that MTBF includes. Reconcile the two by aligning the downtime taxonomy at source — typically in the data model of the historian or MES. ISO 22400-2 provides a downtime taxonomy worth adopting verbatim.

The brownfield reality. Many of the assets reporting MTBF in 2026 are twenty or thirty years old. Their failure data has been collected manually by shift supervisors, often on paper, and uploaded retrospectively. The data is biased toward catastrophic failures (which everyone notices) and against minor stops (which nobody logs). The brownfield IoT legacy machine connectivity playbook covers the retrofit pattern — edge gateway plus protocol bridge plus state-tagging — that makes MTBF data trustworthy for older assets. Without it, MTBF is closer to folklore than measurement.

Maintenance KPIs are the difference between predictive maintenance with real ROI and predictive maintenance as a marketing slide. MTBF, MTTR, MDT, and PM compliance together describe the reliability profile of the asset base with enough fidelity to plan capex, allocate technicians, and feed a CMMS-integrated digital twin.

Energy Intensity and Sustainability KPIs

Energy and sustainability KPIs were optional in 2010, useful in 2020, and unavoidable in 2026. The combined pressure of regulatory disclosure (CSRD in Europe, SEC climate rule in the United States, CBAM at the EU border), customer requirements (Scope 3 reporting up the supply chain), and operating cost makes energy intensity a tier-one KPI alongside OEE.

Energy intensity (kWh per good unit) is the foundational energy KPI for manufacturing. Formula: Energy intensity = Total kWh consumed during production / Good units produced. The denominator must be good units, not total units, otherwise scrap is rewarded as throughput. Computed on a per-shift basis with sub-metered electricity data, energy intensity exposes asset-level inefficiency. A 10% increase shift-on-shift with no production change usually means a compressor leak, a chiller short-cycling, or a process running at suboptimal temperature.

Carbon intensity (kg CO2e per good unit) translates energy into emissions. Formula: Carbon intensity = kWh x grid emission factor + on-site fuel emissions, all divided by good units. The grid emission factor varies by region and by hour — Ireland’s grid intensity at 2 a.m. is roughly half what it is at 6 p.m. Time-of-use carbon intensity reveals opportunities to shift load to lower-carbon periods, which CSRD-disclosing operators are increasingly required to demonstrate.

Figure 4 shows the data architecture for energy KPIs: sub-meters at the input layer, normalisation against MES production data, and the four canonical KPIs at the output.

Figure 4: Energy KPI framework — sub-metered inputs normalised against production to produce kWh/unit, CO2e/unit, water/unit, and energy productivity.

Water intensity (litres per good unit) matters for water-intensive processes — semiconductors, food, beverages, paper, textiles, paint. In water-stressed regions it is a license-to-operate KPI, not just an efficiency KPI. The formula mirrors energy intensity: total water consumed during production divided by good units.

Energy productivity (good units per kWh) is the inverse of energy intensity and is more intuitive for boards and investors. “We produced X units per megawatt-hour this quarter, up from Y last quarter” lands better than “our energy intensity fell from 1/Y to 1/X.” Both are the same data; pick the framing that fits the audience.

Compressed air leakage deserves its own KPI in any facility with significant pneumatic load. Compressed air is the most expensive utility per unit of work delivered — typically four to eight times the electricity needed to run an equivalent electric motor — and most plants leak 20-30% of their compressed air through fittings and worn seals. A compressed-air leakage KPI computed weekly from off-shift baseline pressure measurements pays for itself within months of being tracked.

Standards anchoring matters here. ISO 50001 specifies energy management system requirements and the related ISO 50006 standard defines energy performance indicators (EnPIs) and energy baselines. A factory that aligns its energy KPI definitions with ISO 50006 inherits the regression-based normalisation approach — production volume, weather, product mix — that prevents seasonal noise from polluting the trend line. Without normalisation, energy intensity rises every winter as plant heating climbs, and operators chase noise.

The pitfall: per-unit normalisation hides mix shifts. A plant making heavy products in Q1 and light products in Q2 will see energy intensity per unit rise from Q1 to Q2, even with no efficiency change, because Q2’s units use less energy each. Normalise by an equivalent unit (e.g., kg of product, square metres of substrate) for mixed-product lines, or report per-SKU energy intensity.

Connection to grid signals and time-of-use pricing. A 2026-era smart factory does not just measure energy intensity; it shifts load in response to grid carbon and price signals. The OPC UA FX field-level communications standard enables the sub-second coordination required for demand-response participation. Energy KPIs in the dashboard should include grid-signal-aware variants: “kWh per good unit during peak hours” alongside the all-hours average. The gap between them is the demand-shifting opportunity.

Energy is no longer a back-office cost — it is a strategic axis of competitiveness and disclosure. The KPIs need to match.

Selecting and Stacking KPIs

The biggest dashboard mistake is treating KPI selection as an additive process — add a metric whenever someone asks. The result is forty KPIs and zero attention. Selecting and stacking KPIs is the discipline that decides which metrics belong on which dashboard for which audience.

A working selection rule: every KPI on a dashboard must answer a question someone is currently making a decision about. If no decision depends on the KPI, archive it. Figure 5 captures the decision tree I walk operators through to filter their candidate KPI list.

Figure 5: KPI selection decision tree — start with the problem, then choose the metric.

Layer the dashboards by audience and cadence. Operators on the line need real-time, asset-specific KPIs — current OEE, current FPY, current micro-stop tally — refreshing every 30 seconds. Plant managers need shift- and day-aggregated KPIs across the line — OEE rollup, FPY rollup, MTBF, kWh/unit — refreshing every 15 minutes. Corporate operations needs week- and month-aggregated KPIs benchmarked across plants — TEEP, RTY, MTBF, carbon intensity, TRIR — refreshing daily. Plant directors live in the middle layer.

The three-tier stack:

1. Operator dashboard — three to five real-time KPIs per asset, one screen per cell or line, large font visible from ten metres.
2. Plant dashboard — twelve to fifteen aggregated KPIs spanning all five buckets, one screen per plant, comparative against shift target.
3. Enterprise dashboard — eight to ten cross-plant KPIs, executive narrative around variance, monthly review cycle.

Any KPI on more than one tier should appear with the same name and formula across tiers. The dashboard hierarchy decays the moment “OEE” on the plant screen and “OEE” on the enterprise screen mean different things.

Leading vs lagging stacking. Pair each lagging KPI with at least one leading indicator. OEE is lagging; planned-stop adherence and micro-stop count are leading. RTY is lagging; gauge R&R compliance and inspection frequency are leading. MTBF is lagging; PM compliance and lubrication audit are leading. Energy intensity is lagging; compressor cycling pattern is leading. Leading indicators let teams intervene before the lagging KPI moves.

Benchmarks need cohort context. “Our OEE is 68%” means nothing without a peer cohort. Best practice is to benchmark within plant (this line vs identical line), within company (this plant vs similar plant), and against external cohorts (semiconductor fab vs WEF Lighthouse Network publications, automotive line vs VDA reports). Each comparison answers a different question — internal variance, company best practice, industry frontier — and KPIs need to be stacked to enable all three.

Stop reporting averages without distributions. An average OEE of 70% on a line that runs 90% on day shift and 50% on night shift is operationally useless; the average hides the actual problem. Always pair the average with a distribution (P10, P50, P90) or a stacked time series. Distribution-aware KPI reporting is the single biggest upgrade most factories can make in 2026 with zero new data acquisition.

KPI selection done well is a quarterly review, not a permanent state. The metrics worth tracking change as the operation matures, as products mix shifts, and as regulatory disclosure expands. A quarterly KPI review with explicit retire-and-promote decisions keeps the dashboard alive instead of becoming archaeological.

Trade-offs, Gotchas, and What Goes Wrong

Every advanced manufacturing KPI has a failure mode, and most of them I have seen first-hand. Knowing the trade-off in advance is the difference between a KPI driving improvement and a KPI driving gaming behaviour.

OEE drives “make and store” behaviour. Push OEE hard and operators run extra units to keep performance high during periods of low demand. Inventory rises, working capital balloons, obsolescence creeps in. Counter with a “good units to demand” KPI alongside OEE — produce-against-demand, not produce-against-capacity.

FPY can be gamed by relaxing inspection. If inspection criteria tighten when FPY is high and loosen when FPY is under pressure, the KPI moves but quality does not. Lock the inspection standard at gauge-R&R level and protect it from KPI pressure.

MTBF gets gamed by reclassifying failures as planned events. A reliability engineer who needs to hit a quarterly MTBF target can reclassify two genuine failures as “planned condition-based interventions.” MTBF rises with no change in actual reliability. Counter with explicit failure-classification audits and second-line approval for any reclassification.

Energy intensity gets gamed by deferring maintenance. Switching off ventilation on hot days, deferring chiller PM, running compressors at reduced setpoints — all reduce energy consumption short-term and all backfire. Pair energy intensity with PM compliance and ambient-condition normalisation.

Dashboard fatigue. The single most-common KPI failure mode is not a wrong metric but too many metrics. Forty KPIs across three monitors means no one tracks any of them, and the operation drifts. Five to seven KPIs per audience, well-defined and well-instrumented, beats forty mediocre ones every time.

Practical Recommendations

A short set of working rules from rebuilds I have led:

• Anchor every KPI to a standard. ISO 22400 for production and maintenance, ISO 50006 for energy, ISA-95 for the data hierarchy. Standards-anchored KPIs survive vendor change and audit.
• Compute KPIs per good unit, not per total unit. Scrap should be a cost, not free throughput.
• Pair every lagging KPI with a leading one. OEE with planned-stop adherence, RTY with gauge R&R, MTBF with PM compliance, energy with compressor cycling.
• Report distributions, not just averages. P10/P50/P90 or a stacked time series every time.
• Limit dashboards to five to seven KPIs per audience. Anything more is reporting, not management.
• Run a quarterly KPI review that explicitly retires obsolete metrics. Dashboards drift faster than people realise.
• Reconcile OEE-A with MTBF/MTTR. They must be mathematically consistent on the same data; if they are not, the downtime taxonomy is broken.

FAQ

What are the most important advanced manufacturing KPIs in 2026?
The five-bucket framework gives the answer: OEE and TEEP for output, FPY and RTY for quality, MTBF and MTTR for maintenance, kWh per good unit and kg CO2e per good unit for energy, and TRIR plus training hours per FTE for people. That is roughly ten to twelve KPIs in total, which is the right size for a balanced advanced manufacturing dashboard. Anchor them to ISO 22400 and ISO 50006 so the definitions hold across plants and audits.

How is OEE different from TEEP?
OEE measures effectiveness against planned production time — typically scheduled shifts minus breaks and planned maintenance. TEEP measures effectiveness against total calendar time (24 hours a day, seven days a week). The difference is utilisation: TEEP = OEE x Utilisation. A line running OEE 80% on one shift has TEEP around 27%. OEE answers “are we using scheduled time well?” while TEEP answers “could we make more by running more shifts?” Sales and capital-planning conversations need TEEP; shop-floor improvement conversations need OEE.

What is the difference between FPY and yield?
FPY (first-pass yield) is the proportion of units passing a stage on the first attempt, with no rework. Final yield is the proportion of units that are good after rework. The gap between FPY and final yield is the “hidden factory” — the cost of rework that disappears from final yield but still consumes labour, materials, and time. Tracking FPY (and rolled throughput yield across stages) makes the hidden factory visible and gives quality engineers a real target.

How do MTBF and MTTR relate to OEE Availability?
Availability = MTBF / (MTBF + MTTR), and this is the same Availability that feeds OEE. Improving either MTBF (more reliable equipment) or MTTR (faster repair) raises OEE-A. The two are coupled, and the choice of which to attack first depends on the failure mode: if failures are rare but devastating, focus on MTTR; if they are frequent but quick, focus on MTBF. ISO 22400-2 defines the downtime taxonomy that keeps OEE-A and MTBF mathematically consistent on the same data.

Can OEE be greater than 100%?
No, not legitimately. If your dashboard shows OEE above 100% the cause is almost always denominator manipulation — planned production time has been artificially reduced — or ideal cycle time has been set too generously, so performance reads above 100%. World-class OEE is around 85% for discrete manufacturing. Anything above 90% on a sustained basis warrants a definition audit before celebration.

Which standards should manufacturing KPIs follow?
The primary international standard is ISO 22400, which defines thirty-four KPIs for manufacturing operations management spanning production, quality, maintenance, and inventory. For energy, ISO 50006 defines energy performance indicators (EnPIs) within an ISO 50001 energy management system. The data hierarchy underneath the KPIs follows ISA-95. The VDI 5600 series covers MES architecture and KPI computation. Aligning to these standards makes KPIs portable across plants, auditable for regulators, and consistent with the WEF Global Lighthouse Network reference operations.

Further Reading

• The ISO 22400 standard is the authoritative reference for manufacturing operations management KPIs, with Part 2 enumerating thirty-four canonical metrics.
• ISO 50001 and the related ISO 50006 define energy management systems and energy performance indicators for industrial operations.
• The VDI 5600 series covers MES architectures and KPI computation patterns used widely in German and European manufacturing.
• Vorne’s OEE field guide is the most-cited practical primer on OEE definitions, formulas, and benchmarks.
• The WEF Global Lighthouse Network publishes case studies on factories operating at the productivity and sustainability frontier; their KPI disclosures are useful external benchmarks.
• iSixSigma maintains accessible reference articles on FPY, RTY, DPMO, and the sigma-level translations used in Six Sigma programmes.

For more on the surrounding stack, see our pieces on the IIoT landscape and architecture, the unified namespace architecture with HiveMQ and Sparkplug B, the OPC UA FX field-level communications analysis, and the brownfield IoT legacy machine connectivity playbook.