ERP manages the business, MES manages production execution, and SCADA (with PLCs below it) controls the floor in real time. Each holds a different layer of data — and the gaps between them, where the systems don't share data, are the silos that make numbers disagree and AI stall.
Knowing what each holds is useful. Knowing where they don't connect is what matters for data readiness.
The three layers
Manufacturing systems stack in a rough hierarchy, from the machine up to the business:
- SCADA (and PLCs beneath it) — the floor / OT layer. Real-time control and monitoring of equipment: machine state, runtime, faults, sensor readings, cycle times. This is operational technology, closest to the metal.
- MES — the execution layer. What's actually happening in production: work orders, output, quality results, traceability, labor, and the raw inputs to OEE. It sits between the floor and the business — and is meant to bridge them.
- ERP — the business / IT layer. Running the business: orders, inventory, procurement, planning, scheduling, and finance. It's the top-level view, furthest from the machine.
(A QMS often sits alongside, holding quality and compliance records.)
What each system holds
| System | Layer | Manages | Example data | |---|---|---|---| | SCADA / PLC | Floor (OT) | Real-time machine control & monitoring | Runtime, faults, temperatures, cycle times | | MES | Execution | Production execution | Work orders, output, quality, traceability, OEE inputs | | ERP | Business (IT) | Business management | Orders, inventory, cost, planning, procurement |
Each is good at its job. The trouble starts at the boundaries.
The gaps between them — where the problems live
The data problems most manufacturers feel don't sit inside these systems; they sit in the gaps between them:
- The SCADA↔MES gap. Floor reality and production records don't reconcile — actual machine runtime versus what got recorded, for instance — so the numbers diverge from the moment they're captured.
- The MES↔ERP gap. Plant execution and the business view drift apart — what was actually produced versus what ERP believes was produced — which is why operations and finance so often argue over output.
- The OT/IT divide. The deepest gap of all: SCADA (OT) and ERP (IT) were never designed to talk to each other, and while MES is supposed to bridge them, it frequently doesn't fully. The two worlds speak different languages.
The result: each system holds a genuine piece of the truth, but no system holds the whole truth — and the data falling into the gaps becomes dark data. IDC estimates over 80% of manufacturing data is dark (IDC, 2022), and a lot of it is lost precisely at these seams.
Why the gaps matter
These gaps are the data-readiness problem. You can't compute a true plant-wide OEE or OTIF when the floor, execution, and business layers don't reconcile — and you certainly can't run AI, which needs the whole picture across all three. The gaps are why reports disagree, why decisions are made on partial views, and why pilots fail. Having all three systems isn't the same as having connected data; the value is in bridging them.
How to close the gaps
You close the gaps by connecting the layers into one foundation: integrating SCADA, MES, and ERP data, reconciling it, and agreeing on common definitions so a number means the same thing across all three. That's the work of integration and data engineering. The first step is usually a floor data audit to map exactly where the gaps are. Closing them is the leap from Disconnected to Connected on the Data Maturity Model — the foundation everything else depends on.
A real-world example
(Brief composite illustration — not a specific named client.)
A manufacturer's MES and ERP disagreed on output: the plant's execution records showed one production figure, ERP showed another, and reconciling them by hand ate hours every week. The cause was the MES↔ERP gap — the two systems were never properly connected, so each kept its own version. Bridging that gap into one reconciled foundation gave a single production number both operations and finance trusted. The systems hadn't been wrong individually; the gap between them was the problem.
Frequently asked questions
Next steps
Data Readiness Scorecard
Gauge where your data stands before building anything on top of it.
Take the ScorecardDiscovery & Assessment
We map your data gaps, score your AI readiness, and hand you a prioritised next-step roadmap.
See how it worksBook a Discovery Call
See exactly how we'd approach this for your operation. No pitch decks.
Book a Discovery CallSources
- IDC (2022) — >80% of manufacturing data is "dark" / unused, much of it lost in the gaps between ERP, MES, and SCADA where systems don't reconcile.
- IDC (2022) — >80% of manufacturing data is "dark" / unused, much of it lost in the gaps between ERP, MES, and SCADA where systems don't reconcile.