Reports in the Context of OEE

What are OEE reports?

OEE reports are the structured evaluations that turn continuous machine and production data into decisions — shift by shift, week by week, plant by plant. A good OEE report does not just show a number; it tells a production manager where the losses were, why they occurred, and what action is expected before the next review. A bad OEE report shows a single green number, gets printed on Monday morning, and changes nothing on the floor for the rest of the week.

I have spent 25+ years rolling out OEE reporting across four continents — as a Six Sigma Black Belt at Johnson Controls, as MES program manager at Visteon, and now at SYMESTIC. I wrote a book last year called OEE: One Number, Many Lies, and the title was not a marketing choice. It came from watching the same reporting pathology in plant after plant: the reported OEE goes up every quarter, the actual productivity does not move, and nobody on the management team is willing to ask why. This article is about how to build OEE reports that do the opposite — reports that show the honest picture, name the losses, and drive a measurable response.

The five-layer report hierarchy — pick the right report for the right decision

The first thing that separates plants with useful OEE reporting from plants that drown in it is that each layer of the organisation needs a different report. Same underlying data, different aggregation, different latency, different level of detail. Conflating the layers is the most common reason OEE reports end up being useful for nobody.

The single rule I apply in every engagement: each report layer exists only if it drives a decision the layer below cannot make on its own. Reports that do not change any decision at any level are organisational theatre, and they are more common than plants want to admit.

One number, many lies — the four ways OEE reports hide the truth

Here is the inconvenient observation behind the title of my book. In 25+ years of walking into plants and comparing the reported OEE against the raw event data, I have found systematic distortion in roughly nine plants out of ten. The people producing the reports are not dishonest. The reporting system is quietly structured in a way that produces optimistic numbers, and nobody wants to be the one who exposes it. The four patterns below account for almost every distortion I have ever seen.

1. Survivor-bias scope. OEE is calculated only on "scheduled production time" — which is redefined, quietly, to exclude everything inconvenient. Planned downtime grows. Setup is reclassified as non-productive. Breakfast is removed from the denominator. Over three years, the scope shrinks by 20-30 percent and the OEE number drifts upward by 8-12 points. Productivity in the plant has not changed. The definition has.

2. Definition drift. "Availability" in January does not mean the same thing as "availability" in December. A line that used to count micro-stops under 3 minutes now only counts stops over 5 minutes. "Quality" used to include scrap and rework; this year rework is excluded. Every individual change looks reasonable in the meeting where it is decided. The cumulative effect is that the time series is no longer comparable with itself.

3. Rounded-up single numbers. A single OEE figure — 78 % — is reported for an entire plant with 40 heterogeneous lines running 20 different products. The figure is arithmetically defensible and operationally meaningless. The bottleneck line is at 54 %. The flagship line is at 89 %. The average hides both problems. Any OEE report that does not break down by line, by product and by shift is reporting noise, not information.

4. Reverse-engineered targets. The plant commits to 75 % OEE by year-end. As the deadline approaches, the measurement gets "adjusted" to make 75 % achievable. A target that is supposed to drive improvement drives redefinition instead. This is the most corrosive pattern, because it destroys the credibility of the entire OEE programme for years afterwards.

The single test I apply to any OEE report in any plant: can I reconstruct the reported number from the raw event stream, using the same definitions that were used last year, this year, and the year before? If the answer is no, the report is not measuring productivity — it is measuring the reporting system's ability to produce comfortable numbers. A low-but-honest OEE is worth more than a high-but-drifting one, every single time.

The two charts that actually drive action — Pareto and waterfall

Most OEE reports are dense with visualisations that look sophisticated and change no behaviour. In two and a half decades of this work, two chart types have done almost all the heavy lifting when it comes to turning OEE data into shopfloor action.

The loss Pareto. A bar chart of the top 10-15 stop reasons (or quality defect categories) ranked by total time (or cost) over the review period. This is the chart that tells a shift leader, a production manager or a plant manager exactly where to spend the next week's improvement capacity. If the top three stop reasons account for 62 % of total loss, the team works on those three. Not on the top 15. The Pareto principle is 80 years old and still the single most useful tool in factory reporting, and most MES systems render it poorly.

The OEE waterfall. Starting from theoretical production time, the waterfall shows the stepwise loss down to actual good output — planned downtime, unplanned stops, speed losses, quality losses. A waterfall is honest about where the hours went in a way that a single OEE number can never be. It is also the single best chart for management conversations, because it forces the discussion away from the aggregated percentage and toward the specific loss category.

Everything else — time-series charts, availability heatmaps, shift comparisons — is secondary. If the Pareto and the waterfall are done properly and reviewed weekly, the reporting programme is already in the top quartile of what I see in the industry.

The shift report — the most important report, and the one most plants do worst

If I were allowed to keep only one OEE report in a plant and delete the other four, I would keep the shift report. Not the monthly management view, not the quarterly board pack — the shift report. It is the point where the data actually meets the people who can change outcomes within the next 8 hours. And it is the report that plants most consistently get wrong.

A working shift report has four elements and takes under 60 seconds to read. First, the shift's OEE against target, broken down by line. Second, the top three stop reasons by duration with the name of the person investigating. Third, any quality event that crossed a threshold, with status. Fourth, the open actions carried over from the previous shift. That is it. No attachments, no narrative, no fifteen-tab Excel workbook. A shift report that requires ten minutes to read will not be read, and then the purpose of running the shift report has been defeated.

The diagnostic question for any shift report is brutal: did anyone change what they did this shift because of what was in yesterday's shift report? If the answer is no, the report is not working, regardless of how sophisticated it looks.

A real case: Meleghy Automotive — reporting as the scaling mechanism

Meleghy Automotive is an international Tier-1 supplier specialising in complex body components, with plants in Germany, Spain, the Czech Republic and Hungary — forming, joining and coating processes that are heavily dependent on honest loss visibility. The engagement started in the Wilnsdorf plant as a focused OEE and loss-analysis rollout, and it is the clearest illustration in the SYMESTIC portfolio of how standardised reporting becomes the mechanism for scaling improvement across a multi-site organisation.

The technical backbone was OEE capture at the critical process steps across every plant, feeding a common reporting layer that behaved identically across Wilnsdorf, Gera, Brandýs, Bernsbach, Reinsdorf and Miskolc. The bidirectional SAP R/3 integration via ABAP IDoc tied machine cycles back to production orders, which is the single most important detail in any enterprise OEE reporting: without order-level context, the loss numbers cannot be allocated to products or customers, and the report becomes a plant-internal curiosity instead of a business instrument. A parallel bidirectional link into CASQ-it (Böhme & Weihs) triggered quality sampling from the same event stream — one data layer, two operational consequences.

The scaling pattern is the lesson of this case. Because the reporting logic was standardised in the SYMESTIC configuration, Meleghy rolled out from one plant to six in six months — not because the rollout was cheap, but because every new plant inherited a reporting definition that was already proven. No parallel Excel report, no local dialect of "availability," no plant-specific OEE formula that would have made cross-plant comparison impossible. Meleghy's team now scales the solution themselves through the modular catalogue.

The measured outcomes:

• 10 % reduction in stop time, driven by Pareto-based action on the top loss categories visible in the shift report for the first time
• 7 % improvement in output, from sequencing and loss decisions that now had consistent data across plants
• 5 % improvement in availability, through structured analysis of recurring event patterns across the six-plant footprint

The reporting system was not the cause of the improvement — the improvement actions were. But the reporting system was the reason those actions could be identified, prioritised and compared across six plants without translation. That is the strategic value of OEE reports when they are done well.

FAQ

What should a shift report contain?
Four elements, readable in under 60 seconds. Shift OEE vs. target by line, top three stop reasons with the name of the person investigating, any quality event that crossed a threshold, and open actions from the previous shift. Anything more will not be read; anything less will not drive action. Narrative paragraphs, attachments and fifteen-tab Excel workbooks are what shift reports look like in plants where nobody actually uses them.

How often should OEE reports be reviewed?
The honest rule: reviewed at the cadence at which decisions are made at that organisational level. Live boards continuously. Shift reports at every shift handover. Weekly CI reviews on a fixed day with named owners. Monthly management reports within five working days of month-end. Quarterly reviews as part of the regular governance cycle. Reports that are produced but not reviewed at a defined meeting are a waste of the reporting infrastructure.

Why does the reported OEE keep going up while the actual output does not?
Almost always because definitions are drifting. Planned downtime is growing, the scope of "scheduled production time" is shrinking, micro-stops are being redefined as non-countable, or quality is quietly excluding rework. Over 2-3 years this produces 8-12 points of phantom OEE improvement with zero underlying productivity gain. The only cure is strict definition governance with a single written standard that cannot be changed without sign-off at plant-manager level.

Should OEE be reported as a single number per plant?
Not as the primary metric. A single plant-level OEE averages away the information that actually matters — which lines are the bottlenecks, which products run poorly, which shifts underperform. Report OEE by line, by product family and by shift as the working level, and use the plant-average only as a summary sanity check. Any management report that leads with a single plant number is hiding more than it shows.

What is the difference between an OEE report and an OEE dashboard?
A dashboard is continuous and pull-based — it is looked at when someone has a question. A report is periodic and push-based — it lands at a predictable time with a predictable structure and triggers a predictable meeting. Plants need both. Dashboards answer "what is happening now?" Reports answer "what needs to change by next review?" Treating them as interchangeable produces systems that do one job well and the other poorly.

How do we stop our OEE reports from being ignored?
Three things. First, cut the length by half and cut it again — nobody reads long reports in production. Second, tie every report to a specific meeting with named attendees and a specific decision to be made. Third, run a quarterly audit of whether any action was taken as a result of the report; if the answer is no, the report is killed. Organisational discipline around reports is harder than building the reports themselves, and it is where most reporting programmes quietly die.

Can OEE reports be fully automated?
Generation, yes; interpretation, no. The data collection, calculation, Pareto construction and distribution should be fully automated — there is no excuse in 2026 for a shift leader to spend 45 minutes compiling a report manually. Interpretation, however, requires human context: why did the top stop reason spike this week, which countermeasure is realistic, who owns the action. The purpose of automation is to free the reader's time for the decision, not to produce a report nobody has to read.

How does SYMESTIC support OEE reporting?
Automatic capture of availability, performance and quality at sub-second resolution across 15,000+ connected machines in 18 countries; configurable report layers from operator live boards through shift reports to quarterly executive views; standardised definitions that scale identically across sites (the Meleghy pattern); Pareto and waterfall visualisations as first-class report primitives; bidirectional ERP integration (SAP, Infor, proAlpha, Navision, Dynamics) so reports carry order-level context, not just machine-level numbers. See SYMESTIC Production Metrics.

Related: OEE · OEE Software · KPI Dashboard · Six Big Losses · Shopfloor Management · Real-Time Monitoring · MES · SYMESTIC Production Metrics

About the author

Christian Fieg

Head of Sales at SYMESTIC. 25+ years in manufacturing operations — Johnson Controls (Six Sigma Black Belt, SPS engineer, expatriate in Changchun), global MES lead at Johnson Controls Automotive Electronics with 900+ machines across seven countries, Manager Center of Excellence for global MES at Visteon, MES sales at iTAC and Dürr. Author of OEE: One Number, Many Lies (2025). · LinkedIn