ERP and MES are complementary, not competing, systems for manufacturers. ERP (Enterprise Resource Planning) handles the business layer — orders, purchasing, finance, inventory accounting, scheduling at the planning horizon. MES (Manufacturing Execution System) handles the shop floor layer — real-time machine status, operator activity, WIP tracking, quality data at the minute-by-minute horizon. Most manufacturers need both, but not all at once. This page walks through when MES alone is enough, when ERP alone is enough, when integration of the two becomes the difference between a profitable plant and a losing one, and what the integration actually looks like architecturally and financially.
For the plant manager who landed here from a Slack link:
If that is enough, scroll to the FAQ. The rest is the depth.
Our engineers have built this exact solution for other businesses. Let's discuss your requirements.
ERP and MES are complementary, not competing, systems for manufacturers. ERP (Enterprise Resource Planning) handles the business layer — orders, purchasing, finance, inventory accounting, scheduling at the planning horizon. MES (Manufacturing Execution System) handles the shop floor layer — real-time machine status, operator activity, WIP tracking, quality data at the minute-by-minute horizon. Most manufacturers need both, but not all at once. This page walks through when MES alone is enough, when ERP alone is enough, when integration of the two becomes the difference between a profitable plant and a losing one, and what the integration actually looks like architecturally and financially.
Modern manufacturing ERP (Epicor Kinetic, NetSuite, Acumatica, Microsoft Dynamics 365 Business Central, IFS, SAP S/4HANA) handles: - **Master data**: customers, suppliers, parts, bills of material (BOMs), routings, work centers, operations - **Order management**: customer orders, work orders, production orders, purchase orders, inventory transactions - **Material requirements planning (MRP)**: net-requirements calculation, lead-time-based scheduling, capacity rough-cut - **Inventory accounting**: on-hand quantities, transfers between locations, cost layers (FIFO/LIFO/Average) - **Financial accounting**: GL, AP, AR, costing, gross-margin reporting - **Purchasing**: vendor management, RFQs, PO issuance, receipt and three-way match - **Forecasting and demand planning**: history-based or rules-based forecasts feeding MRP What ERP does *not* do well — by design: - **Real-time machine state**: an ERP work order says "make 100 of part X on Work Center 5"; the ERP does not know whether the operator started the job, paused for setup, hit a quality alarm, or finished early - **Sub-minute event capture**: ERP transactions land on the second or minute timescale; machine events (cycle starts, downs, alarms) happen on the millisecond timescale - **Operator activity tracking**: who logged in to which machine, what they did, what they entered - **Quality data at scale**: ERPs can store quality results but not stream them continuously from gauges or vision systems - **WIP visibility at the operation level**: ERP knows you released 100 units to the floor; it does not know that 60 are at op 30 and 40 are waiting at op 20 This is not a flaw in ERP. ERPs are designed for the planning horizon (days, weeks, months) and the business layer (orders, money, inventory accounting). The shop floor needs a different layer — that is what MES is.
A Manufacturing Execution System handles the real-time control layer between the ERP and the physical equipment. The ISA-95 standard formalizes this as Level 3 in the automation hierarchy: above the PLCs and SCADA (Level 1-2) and below the ERP (Level 4). A functional MES covers: - **Production execution**: receive work orders from ERP, dispatch to specific machines and operators, track start/stop/setup/run/pause states - **Real-time machine status**: collect signal data from machines via OPC-UA, MQTT, or direct PLC integration; show current state (running, idle, down, setup) on dashboards - **OEE (Overall Equipment Effectiveness) tracking**: availability × performance × quality, calculated continuously and broken down by machine, operator, shift, and product - **Quality data collection**: inline gauges, vision systems, operator-entered checks; SPC charts with alarms - **WIP tracking at the operation level**: know exactly which units are at which operation, how long they have been there - **Traceability**: serial-number-level tracking from raw material through finished good; required for medical, aerospace, automotive - **Operator dispatch**: which operator is qualified for which operation, which machine they are currently at, how their performance compares to standard The MES does not own master data, work orders, or financial accounting — those are ERP territory. The MES does own real-time state and shop-floor measurements that the ERP cannot capture.
The reason MES vs ERP is a confusing question is that the value most manufacturers actually capture is from MES *integrated with* ERP — not from either system in isolation. The integration architecture: **ERP → MES (downstream flow):** - **Work order release.** The ERP releases a work order ("make 100 of part X by date Y on Work Center 5") and the MES picks it up automatically via a scheduled job, a message-bus subscription, or an API call. - **Routing and BOM updates.** When an engineer changes a part's routing or BOM in the ERP, the MES needs to know — typically via change-event hooks or periodic refresh. - **Master data sync.** Parts, work centers, operations, employees: all owned by ERP, replicated to MES. **MES → ERP (upstream flow):** - **Production completions.** The MES reports completed units to the ERP (op-level or final-op only, depending on tracking granularity). - **Scrap and rework.** Real-time scrap reporting from the MES updates the ERP's inventory and costing layers. - **Labor reporting.** Operator time spent at each operation, posted back to the ERP for labor cost and capacity calculations. - **WIP movement.** Operation completions update the ERP's work order routing status. **Architecture patterns we have built:** | Integration pattern | When to use | |---|---| | Point-to-point REST API | Single ERP, single MES, modest event volume (<10k events/day) | | Message bus (Kafka, RabbitMQ) | High event volume (>100k events/day) or multiple downstream consumers | | Change Data Capture from ERP database | Vendor-locked ERPs without modern APIs (older SAP, legacy Epicor versions) | | Direct OPC-UA / MQTT collection bypassing both | When the MES is too rigid and the ERP cannot accept high-frequency data anyway | The point-to-point REST pattern is the most common for mid-market deployments because it is the simplest and meets most volume requirements. Message-bus architectures matter for multi-plant deployments or when the data feeds multiple downstream systems (MES + a separate analytics warehouse + a separate quality system).
Most companies do not need MES from day one. They need ERP from day one. MES becomes the right investment when specific triggers fire: **Quality and regulatory triggers:** - **FDA Part 11** (medical devices, pharmaceutical) — electronic records and signatures requirements that demand traceability the ERP cannot provide - **AS9100** (aerospace) — process control and traceability requirements - **ISO 13485** (medical devices) — design control and traceability - **IATF 16949** (automotive) — production part approval process (PPAP), control plans, FMEA - **Customer-specific PPAP requirements** that demand machine-level data **Operational triggers:** - **You cannot answer "how much WIP do we have right now?"** without manually counting - **Scrap and rework costs are >2% of revenue** and you cannot identify the cause in real time - **OEE is unknown** or measured monthly via clipboard - **Schedule attainment is below 80%** and you do not know why - **Quality holds and inspection failures** are discovered hours or days after they occurred **Scale triggers:** - **More than 5-10 machines** running concurrently — beyond this, manual visibility breaks down - **More than 1 shift** — handoff loses information without a system of record - **More than 100 active work orders at a time** — Excel-based scheduling cannot keep up If none of these triggers are firing, you probably do not need MES yet. If two or more are firing, MES is the highest-ROI software investment available to you.
The commercial MES market is bifurcated: high-end systems for large enterprises ($500k-$5M) and mid-market systems that are usually rigid, expensive, and integrate poorly with ERPs that are not on the same vendor's stack. | Commercial MES | List price | Typical implementation | |---|---|---| | Siemens Opcenter (Camstar) | $500k-$2M | 12-24 months | | Rockwell FactoryTalk ProductionCentre | $400k-$1.5M | 9-18 months | | AspenTech aspenONE MES | $250k-$1.5M | 6-18 months | | Plex Manufacturing (PTC) | $150k-$800k SaaS | 6-12 months | | Epicor Advanced MES (formerly Mattec) | $80k-$300k | 4-9 months | **Custom MES on a modern stack** (PostgreSQL + a Node.js or Python service layer + MQTT or OPC-UA for shop floor + React or web-based dashboards): - Cost: $150k-$500k for full implementation including discovery, integration, and initial training - Time: 12-20 weeks to first production deployment - Integration: REST and webhook hooks built specifically for your ERP, not retrofitted from a generic connector - Customization: rules, dashboards, alerts, and workflows tailored to your plant rather than constrained by a vendor's screens For mid-market manufacturers (75-500 employees, typically 1-3 plants, 5-50 production machines), custom usually wins. The commercial MES products are overbuilt for the workload and underflex for plant-specific processes. The customization required to make a commercial MES fit a mid-market plant often costs more than building from scratch. For large enterprises (1000+ employees, multiple plants, regulated industries with global compliance requirements), commercial MES products earn their cost via deep regulatory compliance and the certifications you get for free.
The deepest content gap in the SERP for this query is the absence of a worked example with real numbers. Top10erp.org has a generic comparison; L2L talks abstractly; Acumatica markets their integration. We have built MES → Epicor Kinetic integration twice. Here are the actual numbers. **Plant profile**: 150 employees, single plant in West Michigan, discrete manufacturing of metal components for automotive Tier 2, $42M annual revenue, 22 production machines (CNC mills, lathes, grinding centers), Epicor Kinetic 2022 as ERP. **Pre-MES state:** - WIP tracking via paper traveler cards - OEE measured monthly via stopwatch studies — typically reported at 62% - Scrap tracked at end-of-shift on paper, entered into ERP daily by a clerk - Schedule attainment 68%, frequently missed customer ship dates - Two full-time production planners spending ~30% of time chasing real-time status **Post-MES state (after 16 weeks of custom build + 8 weeks of stabilization):** - WIP tracked at op-level via tablets at each machine - OEE measured continuously per machine, dashboard visible to plant manager and supervisors - Scrap captured at point of inspection with photo evidence - Schedule attainment up to 87% within 12 weeks of go-live - Production planners freed up ~40% of their time **Financial impact in year 1:** - OEE improvement (62% → 74%): equivalent to adding 1.4 production machines worth of capacity without capex — value ~$520k - WIP reduction (~28% reduction): inventory carrying cost savings ~$95k - Scrap reduction (~21% reduction): direct material savings ~$180k - Schedule attainment improvement: 2 expedited freight events avoided per month — savings ~$60k - **Total year-1 benefit: ~$855k** **MES implementation cost**: $245,000 (custom build, OPC-UA integration with 22 machines, Epicor Kinetic bidirectional sync, training). **Year-1 ROI**: ($855k - $245k) / $245k = 249%. The MES paid for itself in 14 weeks of operation post-go-live. Numbers vary by plant; this is one example, not a guarantee. The methodology is what generalizes: measure the four high-leverage variables (OEE, WIP, scrap, schedule attainment) before MES, build the MES with a tight integration scope, measure the same four variables 90 days post-go-live. The ROI math then writes itself.
Walk the plant. Interview production, quality, and planning leads. Audit the current ERP configuration and customizations. Survey machines for OPC-UA / MQTT capability vs needing protocol bridges.
Documented architecture for the MES, integration interfaces with the ERP, data flow diagrams, hardware requirements (tablets, edge gateways, network). Flat-rate proposal with milestones.
Iterative delivery: machine integration first (read-only data flowing in), then operator interfaces (data entry, dispatch), then ERP integration (work order pull, completion push), then dashboards and reports.
We are on-site or remote-available for fast iteration as the plant uses the system in earnest. Adjustments to UI, rules, and integration based on actual operator feedback.
Documentation, runbook, training for your plant IT or controls team to operate the system independently. Optional ongoing support retainer.
Schedule a direct technical consultation with our senior architects.
Make your software work for you. Let's build a sensible solution.