Almost every manufacturing company has an ERP system. Almost every one has an MES - or at least a production data acquisition system. And almost everyone has an Excel spreadsheet, an email attachment or a manual export/import process somewhere in between, which costs hours every day and generates errors every month.
This is called media discontinuity. And it's not a marginal technical problem. It is the reason why quality data in the IATF audit is not available without gaps. That a callback takes three weeks instead of three hours. That the quality indicators in the ERP do not match the real values from production - because there is a manual transmission layer between recording and posting that delays, falsifies and omits.
This article explains what MES/ERP integration actually means for end-to-end quality management: which data flows are required between the systems, which architecture patterns work, which interface protocols are used - and how integration is implemented step by step without jeopardizing ongoing operations.
THE MOST IMPORTANT FACTS IN BRIEF
|
BRIEFLY SUMMARIZED
|
What a media disruption is - and what it actually costs
A media discontinuity refers to the transfer of data between two systems without an automatic, structured interface. Instead, the data transfer is bridged by manual steps: export to a file, transfer by email, manual input into the target system.
This is manageable in a single transfer. In a manufacturing company with hundreds of daily quality data points that need to flow between four to six systems, it is structurally unmanageable - regardless of the diligence of the employees involved.
Four typical media breaks in the quality data flow in production
| From | To | Media disruption | What is lost | Concrete risk |
|---|---|---|---|---|
| MES / machine control | ERP quality module | Manual export, Excel transfer or e-mail | Component-specific process context is missing, time stamps are lost, defect classifications have to be reassigned manually | Only reject figures are recorded in the ERP - without cause of defect and without component reference. Fault analysis and cost allocation at component level are not possible. |
| Inline measuring system / test bench | QMS | Manual export, Excel transfer or e-mail | Measured values are exported as PDF, structured data is lost, manual transfer is prone to errors | Test results are available in the QMS without a machine-readable link to the production process. SPC and trend analysis are limited or impossible. |
| QMS (complaint recorded) | ERP (supplier evaluation) | Manual export, Excel transfer or e-mail | Complaint data is transferred manually, there is a time delay, relevant data is partially lost | The supplier evaluation is incomplete and delayed. Automatic blocks and escalations take effect too late or not at all. |
| ERP (production order released) | MES (order starts) | Manual export, Excel transfer or e-mail | Quality specifications from the ERP control plan are not transferred automatically, inspection characteristics are maintained manually, version conflicts arise | The MES works with outdated quality specifications. Inspections are carried out according to incorrect tolerances or outdated specifications. |
The total costs of a media breach: Direct costs: Personnel costs for manual transfer (typically 2-5 hours per day in a medium-sized manufacturing company). Indirect costs: Delayed reaction to quality problems, decisions based on outdated data, increased audit risk due to gaps in the documentation chain.
The media break is the invisible quality barrier. Not the machine, not the worker - but the moment when a system change is bridged by an Excel file. This is where quality data loses its context. And without context, data is no basis for decision-making.
- Korbinian Hermann CEO, CSP Intelligence GmbH
The three system worlds: What MES, ERP and QMS each have
Before data flows can be planned, it must be clear which system primarily owns and manages which data. This separation of responsibilities is the basis for a clean integration architecture.
| Data type | MES Manufacturing Execution System | ERP Enterprise Resource Planning | QMS Quality management system | MDA/BDE Machine/operating data acquisition |
|---|---|---|---|---|
| Master data | Machine master dataWork plan versionsTool management | Material masterParts listsSupplier masterCustomer master | Test equipment masterStandard referencesSupplier qualification | Sensor configurationLimit value definitions |
| Transaction data | Production order statusConfirmations per operationMachine timesScrap / rework per operation | Production ordersGoods movementsOrdersInvoices | Test protocolsDeviation reportsCalibration protocolsCAPA measures | Real-time sensor dataAlarm eventsOEE raw data |
| Quality data | Process parameters per componentCurve progressionsOnline inspection resultsSPC raw data | Quality master (characteristics)Inspection plans (specifications)Complaint managementBlocked stock postings | Statistical evaluationsQuality certificatesAudit reportsKPIs / quality indicators | Machine statusAnomaly eventsShift logs |
|
THE INTEGRATION PROBLEM AT THE CORE No single system has all quality-relevant data. MES has the process parameters - but not the quality specifications from the ERP control plan. ERP has the quality specifications - but not the actual measurement results from the MES. QMS has the inspection results - but not the component-specific production context from the MES. Consistent quality management only comes about when all three worlds talk to each other - automatically, in a structured way, in real time. |
Critical data flows: What needs to flow where - and why?
Not every data flow between MES, ERP and QMS is equally critical. The following matrix shows the data flows that must not be missing for consistent quality management - with reasons and consequences if they are missing.
|
Data flow |
Direction |
Frequency |
Why critical |
Consequence of media break |
|---|---|---|---|---|
|
Production order with quality specifications |
ERP → MES |
With order release |
MES must know which inspection characteristics and tolerances apply to this order |
MES works with outdated or incorrect tolerances - missing parts are rated as good |
|
Production message with rejects/rework |
MES → ERP |
Real time or end of shift |
ERP requires actual quantities for material booking, cost calculation and delivery capability |
ERP shows incorrect stock levels - delivery commitments and purchasing are based on incorrect figures |
|
Accurate component inspection results |
MES/QMS → ERP |
After test completion |
ERP goods receipt and vendor evaluation require inspection result as basis for posting |
Goods receipt is posted without quality status - blocked warehouse logic does not work |
|
Quality key figures (PPM, RFT, scrap rate) |
QMS → ERP |
Daily / weekly |
Controlling and management control on the basis of ERP quality KPIs |
KPIs are delayed by weeks and incomplete - management decisions based on old data |
|
Complaints to supplier base |
QMS → ERP |
When recording complaints |
ERP supplier evaluation must be updated automatically - basis for purchasing decisions |
Supplier evaluation months old - bad suppliers remain active, escalation takes place too late |
|
Calibration status of test equipment |
QMS → MES |
During calibration process |
MES must not release test equipment with expired calibration status |
Test performed with non-calibrated measuring equipment - IATF finding, test result invalid |
|
SPC alarms and quality warnings |
MES → ERP/QMS |
Real-time |
Quality deviations must be immediately included in escalation and correction processes |
Quality problem is only noticed during manual evaluation - daily to daily delay |
The three integration architectures: Which one suits which company?
There is no universally correct integration architecture. The choice depends on the number and heterogeneity of the systems, the existing IT infrastructure and the company's medium to long-term digitalization goals.
| Architecture pattern | Brief description | Advantages | Disadvantages | Suitable for |
|---|---|---|---|---|
| Point-to-point | Direct coupling of individual systems | Simple with few systems, no middleware effort, quick to implement, full control per interface | Scales poorly, many individual sources of error, high adaptation effort for changes, no central monitoring | Companies with 2-3 systems and a clearly defined integration task |
| Hub-and-spoke | Central integration server (ESB / middleware) | Good scalability, central monitoring, central transformation logic, new systems easier to integrate | Medium complexity and costs, hub as a critical node, logic in the hub can become complex | Medium-sized companies with 4-8 systems and a heterogeneous IT landscape |
| Unified platform | Common database / single source of truth | High data consistency, no classic synchronization, simple cross-system evaluation, good basis for analytics and AI | Highest effort, system change often necessary, greater provider dependency, long-term project | Companies with a comprehensive digitalization strategy and willingness to modernize |
Interface protocols: OPC UA, REST API, database integration in comparison
The choice of communication protocol determines the latency, security, maintenance effort and future viability of the integration. Three protocol classes have become established in modern production integration, each of which has different strengths depending on the area of application.
|
Protocol |
Direction |
Latency |
Field of application |
security |
Recommendation |
|---|---|---|---|---|---|
|
OPC-UA |
Bidirectional |
< 100 ms (real-time possible) |
MES ↔ machine control, MES ↔ MDA |
High (integrated, certificate-based) |
✓ First choice MES layer |
|
REST API |
Bidirectional |
100 ms - 5 s |
MES ↔ ERP, QMS ↔ ERP, cloud integration |
Medium-High (HTTPS, OAuth) |
✓ Standard for ERP integration |
|
SOAP/WebService |
Bidirectional |
500 ms - several s |
Legacy ERP connections (SAP older) |
Medium (WS-Security) |
⚠ Only for legacy systems |
|
Database integration |
Bidirectional |
Seconds-minutes (depending on polling) |
Direct DB coupling of the same technology |
Low (DB access data) |
Only as an interim solution |
|
File import (CSV/XML) |
Unidirectional |
Minute-hours (batch) |
Legacy without API, manual processes |
Very low |
✗ Avoid - media discontinuity remains |
|
Message queue (MQTT, AMQP) |
Asynchronous |
< 1 s (real time) |
IoT sensor data, machine messages |
Medium-high |
✓ For high data volumes, real-time |
Recommendation for MES/ERP quality integration: OPC-UA for MES-side machine coupling and real-time process data, REST-API for ERP-side business logic integration (orders, quality results, master data synchronization). This combination covers more than 90% of typical integration requirements in the manufacturing industry.
Master data as the foundation for integration: the underestimated risk
The most common cause of failed MES/ERP integration projects is not the interface technology. It is the master data. If material numbers in the ERP do not match part numbers in the MES, if production versions are not synchronized, if supplier numbers have different number ranges - then the most perfect interface quickly transfers incorrect data everywhere.
The four master data risk areas
|
Master data area |
Typical problem |
Consequence of integration |
Solution before integration start |
|---|---|---|---|
|
Material master / part numbers |
ERP material number ≠ MES part number - historically grown different number ranges |
Each data record must be mapped manually - integration fails with unknown numbers |
Create mapping table or define common key (e.g. drawing number) |
|
Production/routing plans |
ERP has version 7 of the routing, MES still works with version 5 - manual release never transferred to MES |
Quality specifications in MES are outdated - Check for incorrect tolerances |
Version synchronization as automatic data flow ERP → MES for routing release |
|
Inspection characteristic definitions |
QMS has different characteristic names than ERP control plan - historically developed in different teams |
Inspection results from QMS cannot be automatically assigned to ERP quality characteristics |
Harmonization of characteristic definitions before integration start - mapping table if necessary |
|
Cost centers / production centers |
Cost centers in ERP and work center identifiers in MES do not match |
Production confirmations cannot be automatically posted to the exact cost center in the ERP |
Uniform cost center key definition as a common data field in MES and ERP |
|
MASTER DATA CHECKLIST BEFORE EACH INTEGRATION PROJECT 1. which number ranges are used in which system? Where are there deviations? 2. is there a common, cross-system component key - or does a mapping have to be created? 3. what is the version history of the routings - and is the current ERP version also the active MES version? 4. are all inspection characteristics identically named and defined in the QMS and ERP control plan? 5. how will master data changes in the ERP be automatically propagated to the MES in future? No integration project without a master data audit. These questions save months of debugging after go-live. |
Integration quickly makes bad data bad everywhere. If you start with a master data chaos, you will have a well-networked master data chaos after integration. The quality of the database must be a priority before the first integration project.
-Amadeus Lederle Chief Technology Evangelist, CSP Intelligence GmbH
The 4-phase implementation plan for MES/ERP quality integration
Integration is not a big bang project. Any attempt to link all systems simultaneously will fail due to the complexity of simultaneous changes in running systems. The proven approach: phase by phase, starting with the highest pain point, with clearly defined acceptance criteria for each phase.
| Phase | Period | Goal | Result |
|---|---|---|---|
| 1. as-is analysis and mapping | 2-4 weeks | Create transparency regarding system landscape, media disruptions and costs | As-is analysis, prioritized integration measures, documented master data gaps |
| 2. harmonization and architecture | 4-8 weeks | Clean up master data and define target architecture | Harmonized master data, defined architecture, pilot scope decided |
| 3. pilot integration | 6-12 weeks | Introduce the first automated data flow productively and validated | Validated, monitored live data flow without manual parallel process |
| 4. scaling | Ongoing, per data flow 4-8 weeks | Gradually integrate additional systems and data flows | Fully integrated system landscape with automatic quality data flow |
Frequently asked questions about MES/ERP integration in quality management
What is the difference between MES and ERP in quality management?
MES (Manufacturing Execution System) controls and documents ongoing production in real time: process parameters, machine feedback, component-specific quality measurements. ERP (Enterprise Resource Planning) plans and manages the business processes: Production orders, material bookings, supplier management, key quality figures for controlling. Both are needed in quality management: the MES supplies the quality facts from production, the ERP provides the quality specifications and receives the results for business management.
What is a media disruption in production?
A media break refers to the transition of data between two systems without an automatic interface. Typical media breaks in manufacturing companies: Inspection results are exported as a PDF file and manually entered into the QMS. Production messages from the MES are imported daily into the ERP via an Excel file. Quality indicators are compiled manually from various systems and entered into reporting. Every media break costs time, generates errors and delays the ability to react to quality problems.
What data needs to be exchanged between MES and ERP for quality management?
The critical data flows between MES and ERP for quality management: ERP → MES: Production orders with quality specifications and inspection characteristics so that MES knows the correct tolerances. MES → ERP: Production messages with rejects, rework and quality status per component so that ERP can post correct stocks and costs. QMS → ERP: Complaint data for supplier evaluation, inspection results for incoming goods quality status. ERP → MES: Calibration status of test equipment so that MES blocks non-calibrated measuring equipment.
What does an MES/ERP integration cost?
The costs of an MES/ERP integration vary greatly depending on the scope, system complexity and integration strategy. As a rough guide for the initial integration of two systems (e.g. MES ↔ SAP): Implementation costs €30,000-150,000 depending on the number of data points, complexity of the transformation and quality of the master data. Ongoing costs for middleware and operation: €5,000-20,000 per year. The payback period is typically 12-24 months, calculated from time savings for manual transfer and improved response time to quality problems.
Which protocol is suitable for MES/ERP integration?
Two protocol classes have established themselves as the standard for MES/ERP integration: OPC-UA for connecting the machine level and the MES - the international industry standard for machine communication, secure, bidirectional, with a built-in semantic model. REST API for ERP-side integration - widely used, flexible, supported by all modern ERP systems (SAP, Microsoft Dynamics, Oracle, proALPHA). Outdated protocols such as direct database integration or file import can be tolerated as a temporary solution, but are not a reliable long-term architecture.
Why do MES/ERP integration projects often fail?
The three most common causes of failed integration projects: Firstly, poor master data quality - if number ranges, versions and definitions do not match between the systems, the interface transfers incorrect data. Secondly, lack of departmental involvement - integration is treated as an IT-only project without quality management and production having defined the data requirements. Thirdly, a big-bang approach - integrating all systems at the same time overtaxes resources and risk management. The solution: master data audit first, pilot integration of a data flow, gradual scaling.
What is the difference between MES/ERP integration and a unified platform?
With MES/ERP integration, the existing systems are retained and connected via interfaces. Data is synchronized between systems - there is no common database, but several synchronized copies. With a unified platform, all systems share a common database. There are no synchronization problems because all systems access the same data set. The unified platform offers the highest data consistency and simplest analysis in the long term, but requires considerable migration and implementation effort.
How long does an MES/ERP integration take?
A pilot integration of a single data flow (e.g. production order ERP → MES) typically takes 6-12 weeks from project start to go-live - provided that the master data is clean and the scope is clearly defined. The complete integration of all relevant data flows between MES, ERP and QMS typically takes 6-18 months, depending on the number of systems, the master data quality and the parallelism of the implementation phases.
