Skip to main content

Manufacturing Industry Performance

Is the industry delivering on the dream — plants that improve every cycle, hit OEE targets, ship on time, scrap nothing, attest their carbon, and keep operators safe — at a cost the operator can absorb?

This page measures the industry. The function-level twin — how a single business gauges its own production — sits at Functional Specification and the Scoreboard.

:::note Heuristic Thresholds Numeric thresholds on this page are agent-estimated starting points calibrated to commonly cited industry ranges (e.g. world-class OEE ≈ 85% per OEE.com industry benchmark). Treat as discussion anchors. Validate against named industry research and plant-level data before locking decisions to them. :::

The Dream​

The dream is a plant that improves every cycle, hits an OEE composite at or near world-class, ships on time without expediting, generates minimal scrap, attests its scope-1+2+3 carbon per unit produced, has zero recordable injuries, and runs on a data plane its owners control — at a cost-per-unit that beats the industry median and improves year over year.

The manufacturing industry is performing well when the gap between that dream and the median plant's reality is shrinking.

The gap is shrinking on real-time visibility (Wave-2 SaaS), on PdM (acoustic + vibration AI), on vision QC (mature models), on energy intensity (sub-metering + demand response). The gap is widening on machine identity (DIDs not deployed), DPP readiness (mandates approaching faster than infrastructure), scope-3 carbon attestation (data quality still poor), and the OT/IT skills gap (operators retiring faster than digital-native technicians arrive).

Paired Gauges — Industry Level​

Every gauge below names a target, a warning, a failure mode, and a decision route. Without a paired failure mode each target is optimism.

G1. OEE composite (industry-wide)​

  • Good: Median plant's OEE composite trending up year-over-year; world-class plants at ≥ 85% heuristic per OEE.com benchmark.
  • Warning: Median plant stuck at ~ 60% heuristic with no improvement over 24 months — the typical industry plateau.
  • Bad: OEE composite never computed in real time on the line; only computed weekly in spreadsheets — the loop isn't closed.
  • Decision route: Wave-2 manufacturing intelligence platform deployed at line level; operator dashboards with real-time composite; daily kaizen reviews against the gauge.

G2. Unplanned downtime as percentage of available production time​

  • Good: ≤ 5% heuristic unplanned downtime — sustained quarter over quarter; PdM coverage on top-N critical assets.
  • Warning: 5–15% heuristic — PdM partially deployed, gaps in coverage.
  • Bad: > 15% heuristic — fire-fighting culture; reactive maintenance dominant.
  • Decision route: Acoustic + vibration AI on rotating equipment; spare-parts kanban tied to PdM signals; MTBF + MTTR tracking on a per-asset basis.

G3. First-pass yield (FPY) by line​

  • Good: ≥ 99% heuristic on stable mature lines with vision QC; ≥ 95% on complex assemblies. Industry benchmark — Six Sigma quality target is 99.99966% defect-free [source: Six Sigma DPMO standard].
  • Warning: First-pass yield stuck below target with rework as the relief valve.
  • Bad: First-pass yield trending down while takt time pressure rises.
  • Decision route: Vision QC on dominant defect classes; SPC at upstream process steps; closed-loop CAPA discipline; recipe-level root-cause analytics.

G4. Schedule adherence (on-time-in-full to customer)​

  • Good: ≥ 95% heuristic OTIF across the customer base; expediting is exceptional, not routine.
  • Warning: OTIF in 85–95% range with regular expedite costs.
  • Bad: OTIF below 85% with chronic expedite mode; customer trust eroded.
  • Decision route: Constraint-based scheduling with real-time WIP visibility; supplier dual-sourcing for critical paths; finite-capacity planning above MRP.

G5. Cost per unit produced (real terms, year-over-year)​

  • Good: Cost per unit trending down faster than CPI on stable products; AI yield + automation absorbing input inflation.
  • Warning: Cost per unit flat in real terms — gains exactly offset by input-cost inflation.
  • Bad: Cost per unit rising in real terms — productivity not keeping up with cost base.
  • Decision route: AI scheduling for energy + labour optimisation; PdM for downtime cost; vision QC for scrap cost; supplier-cost benchmarking for material cost.

G6. Energy intensity (kWh per unit produced)​

  • Good: Sub-metered at machine level; per-batch attribution active; trending down via load shedding + idle suppression + HVAC optimisation.
  • Warning: Energy metered only at plant level; can't attribute to product or process.
  • Bad: Energy spend rising with production; no insight into per-machine consumption.
  • Decision route: Sub-metering campaign on top-N consumers (compressors, HVAC, motors, ovens); demand-response platform integration; energy attribution baked into cost-of-goods.

G7. Carbon intensity (kgCO2e per unit produced, scope 1+2+3)​

  • Good: Scope-1+2 fully attributed; scope-3 ≥ 50% supplier-attested with on-chain evidence; per-unit attestation flowing into DPP. Industry benchmark — CSRD requires scope-3 disclosure for in-scope companies from FY2024–2028 phased rollout [source: EU CSRD].
  • Warning: Scope-1+2 attributed via spreadsheet; scope-3 mostly missing or supplier self-report.
  • Bad: No carbon attribution at all; relying on industry averages for ESG reporting.
  • Decision route: Continuous energy + materials metering; supplier-portal upgrades to ingest attested data; machine-DID + on-chain attestation pilot in the highest-volume product family.

G8. Mean-time-between-failure (MTBF) on critical assets​

  • Good: MTBF tracked per asset; trend stable or improving; PdM platform reading the early-warning signals.
  • Warning: MTBF tracked only retrospectively from work-order data.
  • Bad: MTBF unknown; critical-asset failures treated as random.
  • Decision route: Acoustic + vibration AI on top-N rotating equipment; reliability-centred maintenance program; spare-parts strategy tied to MTBF.

G9. Mean-time-to-repair (MTTR) on critical assets​

  • Good: MTTR trending down via PdM (failure predicted → parts staged → technician dispatched before line stops).
  • Warning: MTTR flat with reactive maintenance pattern dominant.
  • Bad: MTTR rising; technician availability constrained; spare-parts on long lead time.
  • Decision route: PdM platform; spare-parts auto-replenishment tied to predicted failures; technician knowledge-base AI.

G10. Recordable injury rate (TRIR) + lost-time injury rate (LTIFR)​

  • Good: TRIR ≤ 1.0 per 200,000 hours worked; LTIFR trending down; near-miss reporting culture active.
  • Warning: Safety metrics stable but no improvement; near-misses under-reported.
  • Bad: TRIR rising; safety treated as compliance not culture.
  • Decision route: Vision + AI safety monitoring (PPE compliance, exclusion-zone violation); behaviour-based safety program; leadership safety walks; structured near-miss investigation.

G11. Sensor freshness (% machines reporting in last 5 min)​

  • Good: ≥ 99% heuristic of metered assets reporting telemetry in real time; gaps investigated within shift.
  • Warning: 90–99% with intermittent reporting from edge gateways.
  • Bad: < 90% — data plane unreliable; OEE composite cannot be trusted.
  • Decision route: Edge-gateway redundancy; UNS pattern with offline buffering; gateway health monitoring on top of process monitoring.

G12. Data-to-decision latency (minutes from event to action)​

  • Good: ≤ 5 minutes heuristic from event detection to operator action on critical alarms; ≤ 1 minute for safety-critical.
  • Warning: Detection within minutes but decision-action latency in hours due to escalation chains.
  • Bad: Events seen only on next-shift report; gap between detection and action measured in days.
  • Decision route: Operator-screen alarms with embedded decision rights; escalation rules at the gateway; AI-generated action suggestions for known patterns.

G13. Digital product passport readiness​

  • Good: DPP infrastructure live for top-revenue product family; on-chain attestations; QR/NFC carrier on label.
  • Warning: DPP architecture chosen; pilot underway; not yet in production volume.
  • Bad: No DPP infrastructure; ESPR mandate dates (2027–2030, source: EU ESPR) approaching without a plan.
  • Decision route: Adopt open standards (GS1 Digital Link, EPCIS, W3C VCs); pilot in highest-volume regulated category; build supplier-data ingestion path for upstream attestation.

G14. Machine identity coverage (% critical assets with cryptographic DID)​

  • Good: ≥ 80% heuristic of revenue-critical assets have a DID; machine wallets active where machine-to-machine payments are deployed.
  • Warning: Identity pilot active; production deployment slow.
  • Bad: Zero DIDs; cannot enable machine-payable APIs, throughput-as-collateral, or autonomous-cell experiments.
  • Decision route: Pilot machine DIDs on top-N critical assets; choose W3C-compliant DID method; integrate with existing CMMS for asset master data.

G15. OEE coverage (% production lines on real-time OEE platform)​

  • Good: ≥ 90% heuristic of lines on the same OEE platform; cross-line comparisons live.
  • Warning: Pilot lines covered; rollout stalled at scale.
  • Bad: OEE computed weekly in spreadsheets; no real-time platform deployed.
  • Decision route: Wave-2 platform selection (Factbird, Tulip, MachineMetrics, or peer); plant-wide rollout plan with operator training; tie to existing ERP/MES via open data plane.

Industry Aggregate Score​

Aggregate across the fifteen gauges. The industry currently scores:

Gauge clusterIndustry stateTrend
OEE + downtime + yield + scheduleWave-2 platforms closing the gap; gap-to-leaders wideningImproving
Cost + energy + carbonEnergy metering accelerating; carbon attribution lagging mandate datesMixed
Reliability (MTBF + MTTR)PdM is the largest verifiable AI win in mfg in this cycleImproving
Safety (TRIR + LTIFR)Vision-AI safety supplementing behaviour-based programsImproving slowly
Data plane (sensor freshness + latency)UNS pattern adoption growing; legacy MES still the bottleneckMixed
Future-instruments (DPP + DID)Infrastructure trails mandate dates; competitive advantage to early moversTrailing

What Performance Reveals​

The plants that compound year-over-year share three patterns:

  1. They computed OEE in real time at the operator screen before anyone else in their sub-vertical. The composite metric on the screen changed operator behaviour every shift; the rest followed.

  2. They invested in the open data plane before the AI features. Unified namespace + OPC UA + MQTT is the moat. AI agents attach where the data is. Plants that did this five years ago have AI optionality today; plants that didn't are still negotiating vendor SDK access.

  3. They started measuring carbon and machine-identity before the mandate. Not because they were idealistic — because they could see the procurement leverage. Buyers that win the next 5-year contract already have the attestation infrastructure in place.

The plants that didn't compound share the inverse: spreadsheet OEE, vendor-locked data plane, no carbon attribution, no machine identity strategy. The gap doubles every cycle.

Context​

Questions​

  • Which of the fifteen gauges is the single highest-leverage one for your plant — and which is its biggest gap?
  • The aggregate-score table says DPP + DID are trailing. What is your readiness vs your buyer's expectation?
  • If you ran the OEE composite calculation right now, do you trust the inputs enough to act on the output? If not, what fails first — sensor freshness, data plane reliability, or composite formula?
  • Real-time OEE on the operator screen changes operator behaviour. What does the data say about your operators' decisions in the last 30 days?
  • The plants that compound got there by acting before the mandate. Which mandate is closest to biting your industry — and what is the smallest move that gets you ahead of it?