New to the metric, or trying to work out why two plants report wildly different numbers? Start with the OEE software reference, which covers the arithmetic and the definition trap behind it.

The bottom line: When OEE data is delayed, slow production can run for hours before anyone acts. ProAlert pushes live OEE every 10 seconds so supervisors can intervene while there is still time to recover the shift. At RHA Manufacturing, a Toyota Tier 1 automotive supplier, this approach drove OEE from 65% to 89% over 39 months, a 23,000%+ ROI.

Why Most OEE Programs Fall Short

Traditional OEE programs share a structural flaw: the data arrives after the window to act has closed.

End-of-shift data is historical, not actionable.

A machine running at 40% Performance for three hours generates a report entry after the fact. By shift end, the loss is permanent and cannot be recovered.
Spreadsheet OEE depends on manual data collection.

When operators fill in downtime logs from memory, data is rounded, delayed, or skipped entirely. The resulting OEE number is a best guess dressed as a metric.
Disconnected systems produce disconnected OEE.

When your downtime tracker, cycle counter, and scrap log live in separate systems, calculating true OEE requires manual stitching that introduces error at every seam.
Die-driven environments are poorly served.

Standard OEE software sets targets at the machine level. In metal stamping and injection molding, the active tooling determines cycle time and cavity output. The machine number alone tells you nothing meaningful about expected production.

How ProAlert Calculates OEE

ProAlert's OEE engine draws from a unified database that also drives Andon alerts, cycle counting, and scrap entry. Every input is live and automatic.

Availability CORE

Calculated directly from active Downtime (DT) calls. The moment a DT call opens, Availability begins falling. When the call closes, it recovers. No manual entry required at any step.

Performance PREMIUM

Actual cycle counts vs. expected cycles per hour. Data flows from EdgeSense IoT (Internet of Things) devices via GPIO (General Purpose Input/Output) sensors, or from manual production entry.

Quality PREMIUM

Good parts divided by total parts produced. Scrap entries feed the Quality component automatically from the same interface operators use to log rejects. No duplicate entry.

Composite OEE

OEE = (Availability × Performance × Quality) / 10,000. Calculated continuously and broadcast to every connected client every 10 seconds, and also the moment any event changes an input... a scrap entry, a timeline correction, a cycle, a call closing. Live, not cached.

One Performance Formula

Performance is computed in units per cycle on both sides of the ratio, so a multi-cavity asset stops inflating its own number. Rolled-up OEE is a sum of actual over a sum of target, not an average of averages, which is the other arithmetic that quietly flatters a plant.

What a Zero Means Is Your Decision

An installation picks whether a computed zero displays as 100 percent, a dash or a zero, and whether cycles recorded while an asset was marked down count as production. Both default to off. These are exactly the conventions that, left implicit, make two plants' figures incomparable.

Subscription-Gated Tiers: Availability is always included. Performance and Quality are subscription-gated with per-asset expiration date tracking. Assets without active subscriptions do not consume server resources calculating metrics they cannot display. The system stays efficient at scale.

ProAlert production run detail for a hydraulic die press running a hood liner on second shift, showing availability 84.4 percent, performance 91.2 percent, quality 98.7 percent and a composite OEE of 76.0 percent beside the time accounting they divide, 7 hours 15 minutes of planned production, 45 minutes planned downtime, 1 hour 8 minutes unplanned downtime across three calls and 6 hours 7 minutes of operation time, with 558 cycles, 558 units, the frozen targets of 90, 87 and 96 percent, and tables listing the three downtime calls and five scrap records raised during the run — **Every number on this screen can be checked against the numbers beside it.** This is one run on one press, opened after the shift closed, showing the same four figures the board broadcasts every ten seconds while the shift is live. Availability is the 6h 7m of operation time over the 7h 15m of planned production. Performance is 558 cycles against the 100 per hour this die is rated for. Quality is 7 units charged out of 558 made, and the 3 units of supplier-caused material scrap are reported beside them rather than charged, because they are not the plant's defect. The three downtime calls add up to the 1h 8m of unplanned downtime, and the five scrap records add up to the four buckets above them. Nobody has to take the 76.0% on trust.

ProAlert product asset stats list showing ten of 88 rate records, each pairing a machine with a product and carrying its ideal cycles per hour, units per cycle and target OEE (Overall Equipment Effectiveness) percentage — **Performance is a comparison, and this is what it compares against.** Every asset and product pair carries its own ideal cycles per hour, its own units per cycle and its own OEE target, because a line making a dozen rolls an hour and a press making hundreds of bases an hour are not the same machine and should not be held to the same number. This plant keeps 88 of these pairs, ten to a page. Set them once per pair and every Performance figure the platform reports is measured against the rate that pair can actually hold, not against a plant-wide average that flatters the fast cells and punishes the slow ones.

ProAlert unplanned downtime Pareto for one injection press over the week of 31 August to 6 September, with a ranked bar chart above the table: eight bars falling from 214 minutes to 9, the first three drawn in red, a cumulative line climbing to 100 percent against a right hand axis and a dashed 80 percent threshold line across the plot. Below it the same eight failure modes as a table, 660 minutes in total, ranked from a seal or gasket leak at 214 minutes and 32.4 percent through a drive fault, a material shortage, a bearing failure, a sensor fault, an operator error and a lubrication failure to a raw material defect at 9 minutes, each row carrying its share and the running cumulative. — **Live monitoring saves the shift. This is how you stop losing the same hour every week.** One press, one week, 660 minutes of unplanned downtime, ranked by what actually caused it. Three failure modes out of eight account for the first 80%, and the platform marks them in the chart and in the table rather than leaving you to work it out: the cumulative line crosses the dashed 80 percent threshold on the third bar, so fix the hydraulic seal, the drive fault and the material staging and you have addressed four fifths of the loss. The reason codes come off the downtime calls the operators already raised, so nothing here was typed twice or reconstructed from memory at the end of the month. The same week can be ranked by call type instead with one click, and any row opens the individual calls behind it.

Going deeper on Availability: our technical brief How ProAlert Measures OEE Availability documents exactly which inputs are automatic and which are entered, how the downtime clock starts and stops, which time window the calculation runs against, and how an early shift release is accounted for. Includes a worked example and a glossary.

10-Second Live Broadcasting

ProAlert's background OEE service queries all active assets, calculates metrics with the appropriate tier flags enabled, and pushes updates to every connected client on a 10-second cycle. This is a live calculation, not a cached snapshot refreshed on page load.

Broadcast Target	Content	How It Works
Asset Dashboard	Availability, Performance, Quality, Composite OEE	Pushed to the `asset:{assetId}` SignalR group. Any client subscribed to that asset receives the update instantly.
Mobile App	Live OEE tile with target vs. actual display	The .NET MAUI (Multi-platform App UI) mobile app receives the same SignalR push. Greyed tiles indicate unsubscribed tiers.
Andon HUD (Heads-Up Display)	Live asset status with OEE overlay	Shop floor Andon display updates automatically when each 10-second OEE broadcast arrives.
Support Users (Multi-Plant)	All-asset OEE stream	Users subscribed to the global group receive OEE data for every asset across all facilities simultaneously.

Die-Driven OEE Resolution

In metal stamping and injection molding, production targets must follow the active tooling, not the machine. ProAlert resolves OEE parameters automatically from the die or tool selected at run start.

Cavity-Level Cycle Attribution

A 4-cavity die produces 4 parts per press stroke. ProAlert attributes output by cavity count, not as single units. Your Performance metric reflects actual throughput, not raw stroke count.
Per-Cycle Output (PCO) Threshold Resolution

Performance targets seed from the die's PCO configuration. Change the die at the press, and the OEE target updates automatically. No supervisor intervention required to maintain accurate metrics.
OEE Baseline Snapshots

A snapshot captures the OEE state at product selection. Post-shift trend analysis starts from a documented baseline, not a reconstructed estimate.

Five Levels From the Number to the Cause

A score is where the question starts. From the daily management board, every metric drills through five levels to the records underneath it, and each one is reachable rather than being a dead tile.

Level	What it shows
1. Trend	The metric over time, in buckets from year down to shift and hour, with the governing target, warning and critical lines drawn on the chart per bucket.
2. Slice	The bucket broken down by a dimension, optionally stacked by a second one, with each slice stating what it was measured over so a percentage cannot float free of its denominator.
3. Pareto	The drivers ranked, with a cumulative curve, so the two causes worth engineering out separate themselves from the nine that are not.
4. Decomposition	OEE split into Availability, Performance and Quality for that slice, which is the answer to "is this a maintenance problem, a rate problem or a scrap problem".
5. Did the fix hold	The same measure before and after the countermeasure at three ranges, so effectiveness is verified rather than assumed.

Where a level genuinely cannot be shown, ProAlert says which prerequisite is missing instead of returning an error page. And on a published board, a visitor can touch through all five levels without signing in, because a number nobody can interrogate is a number nobody trusts.

Automated Anomaly Detection

ProAlert surfaces performance problems before a DT call is ever placed. Three automated detection conditions alert supervisors while there is still time to act.

Condition	Trigger	Why It Matters
Slow Production	Rolling average cycle time exceeds the configured threshold	A machine running 15% below target rate produces fewer parts per hour without ever triggering a Downtime call. Slow Production surfaces the loss before it compounds.
Production Lull	No cycle activity detected beyond the configured gap period	An idle machine that should be running shows as Running status. Lull detection surfaces the gap before a full shift of lost production accumulates invisibly.
Production Warning	Cycles detected on an asset marked Down or Planned	Highest-priority alert state. A machine running while marked Down means the official record and physical reality are out of sync. This condition requires immediate supervisor attention.

Proven Results: RHA Manufacturing

RHA Manufacturing, a Toyota Tier 1 automotive supplier. Results tracked across 39 months (January 2023 through March 2026) and 21,346 production runs:

Sustained OEE

89%

65% baseline → 89% sustained (39 months)

Scrap Reduction

-65%

8,945 scrap parts (2023) → 309 parts (Q1 2026 pace)

Quality Rate

99.8%

up from 98.79% in 2023; Q1 2026: 99.78%

ROI

23,000%+

RHA Manufacturing, 39 months

How RHA achieved this: The management team exports OEE data every morning and reviews it before the shift starts. Downtime events, scrap incidents, and production gaps are correlated to find root causes. Action items are assigned and resolved that same day. The improvement was not instant... it took 8 to 12 months to see the breakthrough. But it was sustained, and RHA subsequently deployed ProAlert to their Mexico manufacturing facility.

How ProAlert Compares

Typical OEE Tools (Vorne XL, Parsec TrakSYS)

OEE only... no maintenance management included
$5,000–$20,000 per line per year plus hardware
Separate CMMS required: additional $6,000–$15,000/year
Integration between OEE and CMMS: $50,000–$200,000
No video evidence capability

ProAlert

OEE, Andon, CMMS (Computerized Maintenance Management), and Video in one platform
Availability always included; Performance and Quality subscription-gated per asset
EdgeSense IoT device for cycle counting at approximately $500 per unit
No integration cost: single shared database
Proven: 65% to 89% OEE over 39 months at a Tier 1 automotive supplier

For IT and Systems Teams

ProAlert's OEE engine runs on standard, supportable technology with no proprietary middleware dependencies.

Component	Technology	Notes
Real-Time Delivery	SignalR WebSockets (ASP.NET Core 9)	Broadcasts to asset-specific groups every 10 seconds. Falls back to long-polling on restricted networks automatically.
Data Storage	SQL Server + Entity Framework Core	All OEE data in a normalized relational schema. Standard SQL Server backup and disaster recovery applies.
Cycle Counting Input	EdgeSense IoT (Raspberry Pi) or REST API (Representational State Transfer)	Cycles submitted via GPIO sensor on the EdgeSense device, or via the Production REST endpoint for ERP (Enterprise Resource Planning) system integration.
OEE Data Access	470+ REST endpoints	Live OEE snapshots and historical data available for Business Intelligence (BI) and ERP integration. JWT (JSON Web Token) bearer authentication on all endpoints.
Deployment Options	On-premises, LAN-based (recommended)	Windows Server and SQL Server on your plant's local network. Cloud-hosted deployments are available on request for facilities with specific requirements.

See live OEE on your machines.

Book a 30-minute demo... we'll walk through your specific production floor challenges and show you what real-time OEE looks like in practice.

Schedule a Demo