Rockwell Automation’s July 2026 survey of 1,560 manufacturing decision-makers throughout 17 nations discovered that 93% of producers already run MES software program. Solely 28% have it enterprise-wide, simply 23% absolutely built-in throughout ERP, PLM, high quality, and OT.
It is a vendor survey, not impartial analysis – however its personal numbers again the studying. 44% rank integration as the highest MES shopping for requirement, 33% name it their greatest information integration downside.
“MES adoption is now not the hurdle, however enterprise scale is,” mentioned Anthony Murphy of Rockwell. Govindaraju and Putra reached the identical rating a decade earlier, with no stake within the reply.
Useful integration is “probably the most troublesome problem,” they present in a 2016 examine of a Bandung metal plant. Adoption is solved. Integration shouldn’t be.
Manufacturing Execution Techniques Have been Outlined as Configurable in 1997
Low-code MES markets itself as new. It isn’t. MESA Worldwide outlined the manufacturing execution system class in September 1997, in White Paper #6: “Customers can configure this set of performance to satisfy their company and plant aims.”
The identical paper is blunt about why: “the configuration and priorities could fluctuate extensively.” In 1997 vocabulary, configure meant selecting which capabilities a plant deploys, not modeling a workflow in a drag-and-drop builder.
Per-plant variability was by no means a defect low-code found. It was the specification MES shipped in opposition to from the beginning.
What ISA-95 Really Says About Stage 3, and What it Doesn’t
NIST GCR 19-022 (Georgia Tech’s Leon McGinnis) explains why MES will get configured reasonably than shipped completed. Under Stage 3, “management consists of executing predefined operations,” a layer already “moderately well-understood.”
At Stage 3, scheduling is “a particularly troublesome determination downside,” with no precise technique that scales under exponential time. Store flooring automation under Stage 3 is a solved management loop. Stage 3 is a choice downside, found plant by plant, late.
Does ISA-95 truly say SCADA sits at Stage 2? Its personal committee web page, fetched immediately for this piece, by no means locations SCADA or MES at any stage. It describes the usual because the interface “between management capabilities and different enterprise capabilities,” particularly ranges 3 and 4.
NIST says Stage 2, a extensively cited web page says Stage 3, practitioners argue Stage 1 or 2, and ISA-95 itself solutions none of them.
The Bottleneck Sits on the Machine, Not within the App Builder
Rockwell’s personal hole, 93% adoption in opposition to 23% integration, factors straight at machine information integration – measurable earlier than evaluating a single MES function.
On Time Edge’s Kim Burndred splits each asset into two classes: Class 1 machines have native connectivity akin to OPC or MTConnect, Class 2 don’t and want an IIoT gateway. Depend them, he says, “earlier than there’s a rollout plan, and lengthy earlier than there’s even a timeline.”
TU Munich researchers wanted 989 traces of code for information assortment throughout three lab machines, communication layer alone. Dreher Consulting’s schedule delta tracks the identical stock: 4-6 months greenfield, 9-15 months brownfield, and even Dreher’s personal low-code success story nonetheless ran 8 months and 1.5 technical FTE.
Low-code compresses the applying half of a manufacturing facility automation software program undertaking. It leaves the machine-connection half untouched.
Connectivity is Not the Similar as Context
OPC UA was supposed to shut this hole. Does turning on a protocol truly ship that means together with the sign? The OPC Basis says its framework “turns information into info.” In apply, engineers report the mannequin going unused.
Marcus Ilgner, after 18 months with OPC UA, discovered nodes “are simply put into the namespace ns=1” (the generic bucket), as a result of the total mannequin carries overhead that “isn’t certain to yield any ROI within the quick and mid-term.” Wautoma Biotech, an gear OEM, calls the end result “the distinction between offering a cable and offering a map.”
That contextualizing work, not one other protocol, is the job an industrial data management platform is definitely requested to do. The declare isn’t false.
The mannequin can carry semantics. What no protocol does is populate it for you, on any industrial automation platform, machine by machine, plant by plant. Manufacturing information administration is usually that mapping work, not the wire.
The place the Class’s Favourite Numbers Come From
Three final result numbers repeated throughout MES vendor pages hint again cleanly, and none survives the journey. The “45% common discount in manufacturing cycle time” traces to MESA’s White Paper #1, a small group “queried in 1993” – a paper whose personal caveat distributors all the time drop, since outcomes “shall be totally different for every person.”
The identical quantity resurfaces, unchanged, in MESA’s 1997 paper, and nonetheless guarantees real-time manufacturing monitoring positive aspects on vendor pages in 2026. McKinsey revealed two ranges a yr aside: 30-50% downtime discount in 2017, unmethodologized, then 5-15% asset availability in 2018, alongside its personal warning in opposition to treating predictive upkeep “as panacea.”
The “22% on-time supply” and “19.4% internet margin” figures attributed to MESA seem in neither paper anybody cites for them.
No One Has Measured This, Together with the Firm Publishing This Piece
No impartial, managed, multi-plant examine establishes a mean OEE uplift attributable to MES. Why not? Fraunhofer IVV explains it plainly – this materials “often cover(s) in sections like ‘classes realized’… efforts put into the technical answer should not scientifically revealed.”
The one methodologized determine on this analysis, from a NIST temporary constructed on 80 interviews, places the whole addressable alternative at “roughly 3.2% discount within the store flooring value of manufacturing” – an order of magnitude under what good manufacturing platform distributors quote.
That absence of proof applies to each vendor on this class, together with the one publishing this piece. An article quoting a double-digit OEE determine is quoting a vendor, not a examine.
Pace is Not Mechanically an Benefit
Fraunhofer ISI’s 2019 regression on Germany’s manufacturing survey discovered digital applied sciences “didn’t have any statistically vital results on complete issue productiveness.”
Combining robots with digital know-how even brought about “interference, leading to a minimal impact of each applied sciences on productiveness.” The information is from 2012 and covers digitalization broadly, not MES – each information matter.
Hershey’s failed 1999 rollout exhibits what velocity does to an unready course of: three methods compressed onto a 30-month schedule as a substitute of the beneficial 48, testing shortened, go-live timed earlier than the Halloween peak.
Meals Trade Govt stories the identical sample in trendy MES initiatives, with acceptance assessments that “by no means simulate actual changeovers, allergen clean-downs, or recall situations.”
A software that lowers the price of bending manufacturing workflow automation to a plant’s present habits, with out fixing these habits first, solely raises publicity to this failure mode.
A Rollout That Failed for Individuals Causes, Not Technical Ones
gbo datacomp describes a Bavarian agency’s MES rollout, rigorously deliberate technically, that also produced poor information high quality six months in. gbo’s personal prognosis was blunt – “the shift supervisors had by no means been concerned within the conceptualization section.”
The repair wasn’t code. Three supervisors joined as champions and helped design enter kinds, and inside 4 months the information seize price rose from beneath 60% to over 92%, with zero improvement. Even gbo, an MES vendor, admits “software program is the smallest downside.”
Panorama Consulting, an impartial auditor with no stake in both reply, lists the recurring causes as course of design, grasp information, and supervisor buy-in. Change latency isn’t certainly one of them.
Tulip, a low-code MES vendor with each motive to argue in any other case, factors to what occurs after go-live as a substitute: a type change turns into “an IT ticket, a vendor engagement, and a launch cycle… the High quality crew stops asking and begins working across the system.” No impartial supply has measured that declare.
The Variable Price Arguing About
Gartner’s July 2025 Magic Quadrant for Enterprise Low-Code Utility Platforms evaluates twelve distributors, from Appian and Mendix to ServiceNow and Zoho. None builds an MES platform for factories. When an analyst says low-code, it means enterprise app supply. When a plant says it, the phrase means one thing else.
Does that hole make the underlying gradient irrelevant? Not fairly. A 2023 examine in Empirical Software program Engineering (Alamin, Uddin and Malakar) mined roughly 33,000 Stack Overflow posts throughout 38 low-code platforms and located Utility Customization was the most important of 40 clusters, at 30%, as a result of customization “that’s not native to the LCSD platforms turns into troublesome.”
These are enterprise instruments – the plant flooring barely exhibits up on Stack Overflow. Making use of that gradient to MES is this text’s proposal, not the paper’s discovering: work inside what a platform ships natively compresses simply, work that leaves it doesn’t.
A working integrator on PLCtalk made the identical level in 2024: “When you solely have fundamental MES necessities, akin to OEE, these might be scripted immediately in Ignition.
Whether it is full block monitor and hint, you might wish to go on a SepaSoft coaching course.” OEE is arithmetic over tags, track-and-trace a mannequin of fabric id – that hole, not low-code versus code, is the variable truly price arguing about.
One Structure That Takes This Place
Iotellect’s MES module is one instance constructed round that boundary. The seller describes it as “ready-to-use MES modules on high of a completely editable low-code core,” masking planning, OEE, monitor and hint, and real-time monitoring – ISA-95 compliant, deployable cloud, on-premise, or hybrid/edge.
The pitch: shipped Iotellect low-code MES software as manufacturing administration software program for what vegetation already want, an open mannequin beneath for what they don’t.
Every thing above helps the shipped-modules half. The market actually is break up between enterprise low-code and industrial IoT MES, and named modules for named capabilities are a checkable declare.
The editable-core half shouldn’t be the identical form of declare, and the proof right here leans in opposition to it. Alamin’s gradient says leaving what a platform ships natively is the onerous case, not the simple one.
No documented, independently verified case turned up on this analysis – for this vendor or every other – of a low-code platform working full ISA-95 Stage 3 execution at manufacturing scale, its configuration beneath model management and maintainable by plant employees after the integrator leaves.
That case could exist. It wasn’t discovered right here, and the absence must be said, not talked round.
The Information Layer Beneath it
None of that works with out the layer described earlier, the one turning uncooked indicators into modeled information. Iotellect’s broader platform, positioned as an Iotellect industrial IoT platform, connects “40,000+ kinds of units and information sources” via 50+ protocols, on fixed-subscription pricing reasonably than per-device billing – a enterprise truth, not a financial savings declare.
What that protocol depend buys a plant, throughout linked manufacturing facility methods, is a smaller model of the identical downside: connectivity, not robotically context.
The Query a Advertising and marketing Web page Can’t Pre-Reply
ISA-95’s personal scope ensures this downside by no means closes. KTH’s evaluate lists “Administration of configuration” as a Stage 3 assist exercise – versioning the mannequin isn’t a platform nicety, it’s a named a part of what MES is meant to do.
Mike Hadlow named the failure mode in 2012 – a crew’s configuration language ended, years later, “again the place we began… onerous coding every part, besides now in a a lot crappier language.” The actual level got here simply earlier than that line: “at a sure stage of complexity, hard-coding an answer will be the least evil possibility.”
Versionable configuration is the minimal bar. It’s additionally trivially claimable: the reply is a bullet level, not an artifact. What no vendor web page can pre-answer is more durable – present the diff of a genealogy-model change that shipped final quarter. Who reviewed it. What broke. How was it caught.
Editability is the place the clock begins.
It isn’t the place it stops.
