Manufacturing Execution System (MES) Development
Every MES software guide is written by a vendor selling one. This one starts from the plant instead: whether you need an MES at all, whether to deploy a platform or commission custom manufacturing execution system development — and what each route costs before the first line goes live.
- accenture
- coursera
- paysafe
- rezcomm
- staffing future
What does MES software actually do on the line?
MES software manages execution: which work order runs on which machine, which operator signed off each step, what the line actually produced, and why it stopped. Where planning systems decide what should happen, a manufacturing execution system records and controls what does happen — as it happens, not a shift later.
The practical MES system definition is a loop: the plan comes down to the floor as scheduled work orders, the floor reports back completions, scrap, and downtime as they occur, and the numbers the plant manages by are live instead of reconstructed at month-end. Genealogy rides along free — every unit traceable to its materials, machine, and operator, which is what an audit or a recall actually asks for. Just as important is what an MES is not: it is not accounting, not planning, and not a dashboard bolted onto old data — if the numbers on screen are yesterday’s, whatever produced them is not a manufacturing execution system.
Do you need an MES — or just better discipline?
Three symptoms say the spreadsheet era is over. Dispatch chases production updates by phone because no system knows order status. Downtime is discovered a shift late and priced never — the downtime cost calculator puts a number on that hour. And a customer audit turns traceability into three days of folder archaeology.
A fourth symptom hides in plain sight: nobody can state the plant’s OEE without arguing about the inputs. Measure it once with the OEE calculator — if availability, performance, and quality come from three different people’s notebooks, the measurement problem is the MES case.
MES vs ERP: which does your plant need first?
ERP plans; MES executes. The ERP decides what to make, what to buy, and what it should cost. The MES tracks what actually happens on the line — which machine ran, who signed off, why the second shift lost an hour. The deciding question is where your data gap is.
If purchasing, inventory, and costing live in spreadsheets, fix planning first — an MES has nothing to report into, and ERP for manufacturing is the prior project. If the ERP exists but learns what happened a shift late, the gap is on the floor and the MES comes first. Plants that buy both at once usually stall both; sequence by where the numbers are most wrong today.
Should you buy an off-the-shelf MES, deploy a platform, or build custom?
Plant size and process oddity decide it, not the vendor’s demo. A single line running one product family with under ~20 operators does not need an MES yet — a disciplined shift log and a takt board beat software nobody maintains. Don’t hire us for that; come back when the second line breaks the spreadsheet.
| Your situation | Enterprise MES (Opcenter, Plex) | Platform (Intellyx, Ignition) | Custom build |
|---|---|---|---|
| One plant, standard discrete lines, first MES | Overkill — licence outweighs the problem | Best fit. Pilot one line in weeks, expand by line. | Don't — the build costs more than the gap |
| Multi-plant, corporate standardisation mandate | Strong fit — governance, validation, vendor accountability | Workable per plant; harder to mandate globally | Only with a platform core underneath |
| Unusual process (custom routing logic, odd genealogy, legacy machines) | Workarounds at every edge; change orders compound | Platform core + custom modules covers most of it | Justified when the process is the competitive edge |
| Heavily regulated (21 CFR Part 11, aerospace traceability) | Strong — validation packages exist | Achievable; validation effort sits with you | Achievable; budget the validation like a module |
What does an MES cost?
MES pricing punishes big-bang thinking: the honest unit of purchase is one line, not one plant. The bands below reflect a pilot-first path; downtime you already have prices the other side of the ledger — the downtime cost calculator runs that number.
| Cost line | Enterprise MES (Opcenter / Plex) | Platform (Intellyx / Ignition) | Custom build |
|---|---|---|---|
| Licence / subscription | $50,000–$150,000+ per plant/yr | $6,000–$30,000 per plant/yr | None — you own it |
| One-line pilot (8–12 weeks) | $40,000–$120,000 | $8,000–$30,000 | $25,000–$75,000 |
| Plant-wide rollout | 2–4× pilot | 1.5–3× pilot | 2–3× pilot |
| Machine connectivity | $1,000–$2,500 per machine | $400–$1,500 per machine | $400–$1,500 per machine |
| Annual support | 18–22% of licence | From $500/month | From $500/month |
The spread inside each band is mostly machine connectivity — a 2019 line with OPC UA costs a fraction of a 1998 line that speaks nothing. Count your machines and their ages before believing any quote, including one of ours. Budget the pilot as its own decision with its own payback — a plant-wide number quoted before a line has run is a guess wearing a spreadsheet.
The same pilot bands in the major manufacturing hubs
| Market | One-line pilot (platform) | One-line pilot (custom build) | Machine connectivity (per machine) | Monthly support |
|---|---|---|---|---|
| United States (USD) | $8,000 – $30,000 | $25,000 – $75,000 | $400 – $1,500 | From $500/mo |
| India (INR) | ₹3,80,000 – ₹14,30,000 | ₹11,90,000 – ₹35,70,000 | ₹19,000 – ₹71,500 | From ₹23,750/mo |
| Germany (EUR) | €6,900 – €25,800 | €21,500 – €64,500 | €345 – €1,290 | From €430/mo |
| United Kingdom (GBP) | £5,900 – £22,200 | £18,500 – £55,500 | £300 – £1,110 | From £370/mo |
| Mexico (MXN) | MX$149,000 – MX$558,000 | MX$465,000 – MX$1,395,000 | MX$7,400 – MX$27,900 | From MX$9,300/mo |
Converted at 1 USD = ₹95.20 / €0.86 / £0.74 / MX$18.60, 1 September 2026. USD is the contracting currency. India carries regional delivery pricing at roughly half the USD bands; other figures are indicative conversions, not regional pricing.
Which MES platforms are worth evaluating?
Five realistic candidates and a build option, compared on what decides the outcome. Intellyx is Perimattic’s own platform — its row carries the same honest-weakness column as the rest, with every weakness drawn from delivered integrations rather than datasheets.
| Platform | Best fit | Strength | Honest weakness |
|---|---|---|---|
| Intellyx (Perimattic) | SMB–mid-market plants, first MES, pilot-first rollouts | Production, quality, and machine data in one layer; deploys line-by-line; extended by the team that builds it | Younger ecosystem — fewer off-the-shelf connectors than incumbents; gaps closed by development rather than marketplace |
| Ignition (Inductive Automation) | Plants with strong controls engineering in-house | Unlimited licensing, superb PLC connectivity | It is a toolkit, not an MES — you build the MES on it, and that build is the project |
| Siemens Opcenter | Large regulated discrete manufacturers | Deep functionality, validation pedigree | Implementation measured in years and consultants; small plants drown in it |
| Plex (Rockwell) | Automotive suppliers, multi-plant standardisation | Cloud-native, strong quality management | Per-plant pricing climbs; customisation depends on Rockwell partners |
| SAP Digital Manufacturing | Plants already deep in SAP ERP | Native ERP integration | Justifiable only inside the SAP estate; costs assume enterprise budgets |
| Custom build | A process that is itself the competitive edge | Encodes your process exactly — no workarounds | You own the roadmap, which means you fund it too |
Is MES a type of SCADA?
No — they sit at different layers. SCADA watches machines in real time: signals, setpoints, alarms, an operator screen per cell. MES software manages execution across the plant: orders, people, materials, quality, genealogy. The MES sits above SCADA and consumes its data; a plant with excellent SCADA can still be blind at the order level.
The practical test: SCADA can tell you machine 7 stopped at 14:02; only an MES can tell you which order it stalled, what that did to Friday’s shipment, and whether the same fault stopped the same product last month.
SCADA platforms an MES typically sits above and consumes data from:
Does a cloud MES work on a real shop floor?
Yes — if the implementation treats the floor as hostile territory. A cloud based manufacturing execution system works when line terminals buffer locally through dropped connections, barcode scans queue offline and sync on reconnect, and machine data lands through a site gateway rather than each PLC calling the internet.
The question that separates real deployments from demos: “what happens when the network drops mid-shift?” A good answer names local buffering and store-and-forward by design. A bad answer says the Wi-Fi is reliable. Line-side displays are the second test — operators will not wait two seconds for a screen refresh between cycles.
How does an MES connect to machines and existing systems?
Three integration surfaces decide whether the MES reflects the plant or lags it. Machine data arrives via OPC UA, Modbus, or MQTT through a site gateway — industrial IoT development is that discipline. The ERP hand-off moves the plan down and actuals back, so planning stays honest. Quality data flows in from inspection — including camera-based checks — and out to hold rules that stop suspect lots automatically.
Downtime and condition data compound in value once collected: the same machine signals that feed the MES feed failure prediction, which is where unplanned downtime actually starts falling. Where the MES sits in the wider floor stack is mapped on the manufacturing software hub.
How long does a pilot take, and what goes wrong?
One line, 8–12 weeks: connect the machines, model the routings, run the line on MES numbers for a month, and decide plant-wide with data instead of faith. If the OEE and downtime numbers don’t justify expansion, stop at one line — that outcome is cheap, and rarer than the fear of it.
Two failures account for most MES wreckage. Operator rejection: a system that adds clicks to a takt-paced job gets worked around within weeks — the fix is designed-in speed (scan, don’t type; confirm, don’t describe) and a floor champion named before kickoff, not after the stall. And dirty routings: BOMs and routings that disagree with how the line actually runs get faithfully automated, and the MES reports precise nonsense. Walk the line and fix the routings first; cycle times worth trusting come from measured cycle-time data, not the router’s memory. The adoption test is simple to run: time an operator’s MES interaction against the line’s takt — if the confirm takes longer than the takt allows, the design fails before the software does.
Frequently asked questions
What is MES software used for?
MES software manages and records execution on the factory floor: dispatching work orders to machines and operators, capturing completions, scrap, and downtime as they happen, enforcing quality checks, and keeping unit-level traceability. It closes the gap between what planning intended and what the line actually did.
Is SAP an MES system?
SAP's core product is an ERP, not an MES. SAP does sell a separate MES — SAP Digital Manufacturing — which is a distinct product layered under the ERP. Running SAP ERP does not mean you have an MES; execution tracking is a separate purchase or build either way.
What are the most popular MES systems?
Siemens Opcenter and Plex lead among large manufacturers, and Ignition is widely used where controls teams build their own. Popularity is a poor selection filter, though — plant size, process type, and machine ages decide fit, which is what the comparison table above is sorted by.
What is MES software?
MES (Manufacturing Execution System) is the software layer between the shop floor and the ERP. It tracks what's happening on the line in real time: work order status, machine state, operator activity, quality events, batch genealogy. Where the ERP treats production as a scheduled promise, MES treats it as the current reality — the batch running now, the machine that just went down, the operator on shift. Together they give the plant an accurate picture the ERP alone cannot produce.
MES vs ERP — which do I need first?
ERP first if the pain is in orders, inventory, purchasing, or financial close. MES first if the pain is on the line itself: work orders lost between office and floor, machines idle without anyone knowing why, batch traceability failing an audit. Most plants need both eventually, but rolling them simultaneously fails more often than it succeeds — sequence them so the first system is stable before the second starts.
How much does an MES cost?
Pilot bands typically run $8,000-$40,000 for an off-the-shelf platform pilot, $10,000-$50,000 for an open-source platform pilot with custom modules, and $12,000-$120,000 for a bespoke MES pilot depending on machine count and integration depth. A rolling monthly support retainer runs from $500/month. Full-plant rollout scales from the pilot cost roughly linearly with line count, though shared services (identity, master data, reporting) do not repeat per line.
How long does MES implementation take?
A single-line pilot on an off-the-shelf platform runs 8-14 weeks from kick-off to production, including operator training and shift handover. Platform-based pilots with custom modules run 12-20 weeks. Bespoke MES pilots typically run 16-24 weeks. Full-plant rollout after a successful pilot usually adds 6-10 weeks per additional line for platform-based work, less if lines share a common configuration.
What are the main modules of an MES?
The typical core: production execution (work order dispatch, operator confirmation), machine data collection (OEE, downtime reasons, cycle counts), quality management (in-process checks, non-conformance workflow), material tracking (WIP genealogy, lot control), and dispatching (real-time schedule reordering). Advanced modules add SPC, energy monitoring, tool life management, and full traceability against regulatory frameworks like FDA 21 CFR Part 11 or ISO 22000.
Does an MES replace SCADA?
No. SCADA controls machines and processes in real time — it's the safety-critical layer that decides when a valve opens or a servo moves. MES tracks and directs production activity — it decides what to make next, records what was made, and reports what happened. They run alongside each other: SCADA is the machine's nervous system; MES is the plant's operational memory. Modern MES platforms consume SCADA data via OPC UA to enrich production events, but they do not replace the control layer.
Cloud MES vs on-premise MES — which should we choose?
For the transactional layer (work orders, quality records, reporting), cloud is fine — latency doesn't matter when a person clicks a button. For anything the line depends on in real time (dispatch, machine-triggered events, downtime alerts), keep an edge node co-located with the machines and let it sync to a cloud back-end asynchronously. The hybrid pattern — edge for real-time, cloud for analytics and cross-plant reporting — is now the default for greenfield MES deployments.
How does an MES connect to PLCs and existing systems?
Modern PLCs (Siemens S7-1500, Rockwell ControlLogix, Beckhoff TwinCAT) speak OPC UA natively; MES subscribes via an OPC client or through an MQTT/Sparkplug B broker. Older PLCs get a protocol gateway (Kepware, Ignition, Cogent) that translates their native protocol into OPC UA. For legacy ERPs and QMS, REST or message-queue integrations preserve existing workflows while MES becomes the master record for shop-floor events. The rule: instrument the machine without modifying it — retrofits sit outside the safety-certified envelope and add no regulatory burden.
Run the numbers yourself
Machine Downtime Cost
What an hour of unplanned downtime actually costs your line, priced from your own inputs.
Open the Machine Downtime Cost →OEE Calculator
Availability × performance × quality from one shift of real data — the measurement the MES case rests on.
Open the OEE Calculator →Cycle Time Calculator
Measured cycle times to replace the router's memory before routings get encoded.
Open the Cycle Time Calculator →Product Cost Calculator
A first estimate of per-unit cost across materials, labour, and overhead.
Open the Product Cost Calculator →
Start With One Line, Not a Leap
An 8–12 week pilot on your worst line: machines connected, routings modelled, a month of live numbers — then a plant-wide decision made on data. Fixed scope, and if the numbers say stop, we say stop.