In the executive boardroom, the screen displays a flawless Business Intelligence dashboard. Subdued colors, dynamic bar charts, and cards with production metrics updated early this morning.
The corporate IT department completed the project ahead of schedule: it connected the analytical tools directly to the ERP database tables, eliminating the need for manual spreadsheets at the senior management level.
From an aesthetic and executive standpoint, the factory appears to be fully digitized.
However, when the production director goes down to the production floor to understand why one of the main production lines ended the month with a 14% loss in capacity, the answer isn’t on the dashboard.
The ERP dashboard only shows that the Production Order was completed six hours late and that energy consumption was slightly above average. What happened during those six hours remains a complete mystery.
The explanation for this short-sightedness lies in a common structural flaw in the digital transformation of manufacturing: the attempt to directly connect Level 2 (control and automation) to Level 4 (corporate management/ERP), ignoring the existence of Level 3 (manufacturing execution/MES).
BI systems plugged directly into the ERP operate on transactional, static, and lagging data. They were designed to view consolidated accounting events, not to interpret the continuous micro-friction of the physical realities on the factory floor.
By skipping the execution layer, the organization creates dashboards that show the loss figure but remain unable to identify where the margin was lost.
The fundamental reason why corporate BI fails when attempting to analyze the production line lies in the technical disparity between the two worlds of industrial data:
When the technical team attempts to pull data directly from the PLC into the ERP database or integrate BI with corporate tables to analyze efficiency, a conceptual gap emerges:
A high-speed production line doesn’t just stop because of catastrophic breakdowns lasting four hours straight. The biggest thief of installed capacity is micro-downtime lasting 15 to 45 seconds: a misaligned optical sensor, a bottle that gets stuck on the conveyor belt and requires manual intervention by the operator, or a thermal fluctuation that causes the machine to reduce its nominal cycle rate by 20%. The PLC detects each of these events; the ERP lacks the data structure to store them without crashing the relational database.
Since the transactional system does not support millisecond-level granularity, the operation relies on a consolidated report at the end of the shift: the operator manually enters that the line ran for 8 hours and stopped for 40 minutes due to a “mechanical adjustment.” The data feeding the BI is already skewed, simplified, and lacks a physical correlation with how the machine actually operated.
Companies that rely exclusively on BI reports generated from the ERP system face three daily decision-making pitfalls:
Many corporate dashboards are marketed internally as real-time monitoring, but they rely on batch synchronization routines that run in the early morning or after the end of a shift. The director only discovers that the production line lost efficiency on Tuesday at 9:00 a.m. on Wednesday, by which time the defective batch has already been sent to the warehouse and the machine has continued to run out of calibration for two more shifts.
The BI dashboard shows that the unit cost of the SKU increased by 12% compared to the standard technical specifications. However, when clicking on the chart, the system cannot pinpoint the cause: was it the supplier’s raw material that required a slower production speed? Was it the fluctuating temperature in the drying tunnel? Was it the tooling change that took twice as long as expected? The ERP system records the extra cost, but the Level 2 automation system—which stored the reason—did not communicate with the system.
Without an intermediate layer to standardize the variables of heterogeneous equipment, each machine manufacturer delivers its own proprietary supervisory control and data acquisition (SCADA) software. The factory ends up with ten different screens on the shop floor and a central ERP system in the office that does not communicate with any of them in a standardized way.
To eliminate this “dead zone” between the shop floor and the office, the international manufacturing architecture established the ISA-95 model. The standard clearly establishes that Level 2 must not feed Level 4 directly without passing through Level 3: the MES (Manufacturing Execution System) layer.
The MES is neither a substitute for ERP nor a screen display system like SCADA; it is the operational contextualization layer that transforms physical data into business information:
The MES addresses this gap in four fundamental steps:
The electrical signal from the PLC indicating “speed dropped from 120 to 80 ppm” gains business context within the MES: the system links this drop to Production Order No. 45892, the operator logged in at the station, the installed mold, and the specific batch of raw material in use. The data ceases to be an isolated frequency graph and becomes a production cost event.
The MES monitors the PLC’s input and output states. When the line stops counting parts for more than 5 seconds, the system instantly records the micro-downtime and, if necessary, prompts the operator to categorize the reason on the touchscreen terminal if the automation system does not provide a native fault code. No losses slip through the cracks in accounting.
Instead of overloading the corporate database with millions of unnecessary telemetry records, the MES processes events at the execution level and returns to the ERP only the information relevant to accounting governance: actual machine downtime, confirmed good quantities, actual input consumption per cycle, and scrap segregated by cause.
When data analysts and controllers connect the corporate BI system to the MES layer (or its contextualized Historian ), the dashboards gain real analytical depth. The profitability chart by product now allows for a detailed breakdown down to the exact minute the conveyor belt stopped, correlating process parameters with financial losses on the bottom line of the income statement.
Structuring manufacturing governance in accordance with the ISA-95 level separation delivers clear returns for IT and Operations leadership:
Building modern BI dashboards connected only to the ERP is like installing a precision speedometer in a vehicle without connecting it to the wheels: the dashboard has sophisticated lights, but it does not indicate the actual speed of travel.
Artificial intelligence, executive BI, and cost control only deliver useful insights if field data is captured with the frequency, context, and physical accuracy that manufacturing demands.
By implementing the MES layer as the indispensable bridge between the PLC and corporate management, the industry closes the Level 3 blind spot, transforms telemetry noise into financial visibility, and shields its margin against invisible losses.
Is your executive team still trying to understand what’s happening on the shop floor by looking at outdated ERP reports and disconnected spreadsheets?
The PlantSuite MES platform acts as the native Level 3 layer in compliance with the international ISA-95 standard. Integrate PLC telemetry from any manufacturer via the IoT Data Hub, Production Tracking, and Historian modules, contextualizing physical events in real time and delivering reliable, auditable data to your corporate ERP.
👉 Talk to the experts at PlantSuite and find out how to bridge the gap between the shop floor and executive decision-making.