Gartner has refreshed its Magic Quadrant and Critical Capabilities research for Manufacturing Execution Systems, and the usual cycle has already started: vendors are pulling quadrant placements and critical-capabilities scores into sales decks, account teams are leading renewal conversations with “we moved up,” and plant IT teams are getting asked to explain why a system they’ve run for a decade isn’t in the leaders’ box anymore. None of that is new. What’s worth a practitioner’s attention this cycle is what changed inside the methodology — the weighting toward composability, cloud-native architecture, and AI-assisted operations — and whether those categories actually correspond to the problems you’re solving on your floor.
They partly do. They also partly don’t. That gap is where the real work of evaluating an MES vendor happens, and it’s work Gartner’s research, however useful as an input, cannot do for you.
What actually shifted in the methodology
Gartner’s Critical Capabilities reports score vendors against defined use cases — typically something like discrete, process, and hybrid manufacturing profiles — weighted by capability categories. The 2026 refresh leans harder into three areas that weren’t weighted nearly as heavily even a few cycles ago: composability (can you swap or extend modules without a full platform rewrite), cloud-native architecture (multi-tenant SaaS delivery, containerized deployment, elastic scaling), and AI-assisted operations (anomaly detection, generative assistants for root-cause analysis, copilots layered onto historian and quality data).
This tracks with where the major platforms have been investing. Siemens, Rockwell Automation, AVEVA, SAP, Honeywell, and Critical Manufacturing have all pushed public roadmaps toward cloud-hosted MES delivery and embedded AI features over the past several product cycles, so it’s not surprising Gartner’s weighting caught up to vendor R&D spend. The analyst firm is, in effect, scoring the industry on the axis the industry chose to compete on.
Where the weighting actually maps to shop-floor pain
Composability is the one category on this list that genuinely correlates with problems practitioners live with daily. MES implementations fail or stall for years over exactly this issue: a monolithic system where changing a batch record template or adding a new work-order type requires vendor professional services, a change request, and a multi-month wait. A platform built around modular services with documented APIs — ideally exposed through OPC UA or MQTT Sparkplug B on the shop-floor side, and REST/GraphQL on the enterprise side — genuinely reduces the operational drag of running MES long-term. If your last MES project involved a change-control queue that took longer than the change itself, composability is not an abstract Gartner category. It’s your actual pain.
Cloud-native architecture matters, but unevenly, and this is where practitioners should slow down. For a discrete manufacturer running a single site with a modern network backbone, cloud-hosted MES with edge buffering for network interruptions is a reasonable, often lower-friction path to upgrades and multi-site standardization. For a process manufacturer in a regulated industry, or a plant with restrictive network policies, air-gapped security requirements under IEC 62443, or genuinely poor connectivity, cloud-native scoring high on an analyst chart doesn’t change the fact that a hybrid or on-premises deployment may be the only viable option. Gartner’s own reports typically acknowledge this with deployment-model breakdowns — the problem is that a summary quadrant position strips that nuance out, and a sales deck strips it out further.
The category that deserves the most skepticism: AI-assisted operations
This is the one to interrogate hardest. “AI-assisted operations” as a scoring category is currently a mix of things that are genuinely useful (statistical anomaly detection on process data, computer-vision-assisted quality inspection, natural-language query layers over historian data) and things that are largely marketing surface — a chat interface bolted onto existing SPC or downtime data that doesn’t change how an operator actually catches a defect or how a plant engineer actually root-causes a scrap spike. Ask any vendor citing a strong AI score for a specific, demoable workflow: what data feeds it, what’s the false-positive rate in a live environment, and what happens when the model is wrong. If the answer stays at the level of the roadmap slide, treat the score as aspirational, not operational.
Build your own scorecard instead of borrowing theirs
The practical move is to take Gartner’s category list as a starting menu, not a final weighting, and rebuild it against your own constraints. A workable approach:
- Start from your ISA-95 integration reality. Score vendors on how cleanly they integrate with your actual ERP, historian, and quality systems today — not a reference architecture in a datasheet.
- Weight composability by your change velocity. If you add product lines or rework work-order logic often, weight this heavily. If your process is stable and rarely changes, weight it lower.
- Weight cloud-native by your actual network and security posture, not by industry direction of travel. A regulated or air-gapped site should discount this category regardless of where the market is heading.
- Score AI features only on demoed, referenceable workflows in a plant with a comparable process — not roadmap commitments.
- Add a category Gartner doesn’t score well: total cost of ownership over the contract term, including services engagements, module licensing as you scale sites, and the effort of your own team to maintain configuration.
Run each shortlisted vendor through that scorecard with your own weights, and compare the result to where they land on Gartner’s quadrant. When they agree, that’s a useful confirmation. When they diverge — and they will, especially for anyone outside a straightforward high-connectivity discrete manufacturing profile — trust your own scorecard. It’s built on your constraints, not a market-wide average.
What to actually do with the sales deck
When an account team leads with quadrant placement, ask them to walk the Critical Capabilities report’s use-case scoring instead — most publish scores broken out by discrete, process, and hybrid manufacturing profiles, and by deployment model. That breakdown is far more useful than the quadrant graphic, and it’s usually sitting in the same report the vendor is citing from. If they can’t or won’t walk you through the use-case-level detail, that tells you something about how closely their pitch tracks the actual research versus the marketing summary of it.
The Gartner refresh is a legitimate, useful input to a selection or renewal decision. It is not a substitute for scoring vendors against your own floor, your own network, your own change-control pain, and your own contract terms. Read it, use it, and then set it down before you sign anything.
This article was written with the assistance of artificial intelligence. While we aim for accuracy, the information may be incomplete, out of date, or incorrect, and should be independently verified before you rely on it for any decision. It is provided for general information only and does not constitute professional advice.
