For years, IT providers have sold networks the same way: more bandwidth, faster Wi-Fi, better switches, new firewalls, lower cost per port.
But manufacturing executives don't buy networks. They buy uptime, production, quality, and profitability. Once you start looking at the data, that distinction stops being a talking point and starts looking like the whole story.
The Cost of Getting It Wrong
Unplanned downtime is one of the largest controllable drains on manufacturing profitability, and the numbers back that up from every direction. Aberdeen Group's widely cited benchmark puts the average cost of unplanned manufacturing downtime at $260,000 per hour, a figure Siemens and multiple 2025–2026 studies have corroborated — though it varies enormously by sector, from roughly $187,500 an hour in heavy industry to well over $2 million an hour on an automotive line.[1]
A 2025 survey of more than 600 U.S. manufacturing leaders found facilities lose an average of 30 hours of production a month to downtime, and six in ten leaders said these disruptions cost their business more than $250,000 a year.[2] Fluke's 2025 global survey put the number even higher in aggregate: 61% of manufacturers reported unplanned downtime in the past year, adding up to as much as $852 million a week across the sector.[3]
None of that is a "packet loss" problem in the way a network engineer would define it. It's a production problem that happens to have a network at the root of it.
Bars are scaled for readability, not linear — automotive downtime costs are an order of magnitude above the manufacturing average. Source: Aberdeen Group, corroborated by Siemens (2024–2026).
The Missing Link Between IT and OT
Most manufacturing facilities run two separate worlds side by side.
IT (Information Technology) — internet, firewalls, switches, wireless, servers, cloud. Its job is to keep people connected.
OT (Operational Technology) — PLCs, robots, sensors, HMIs, SCADA, industrial automation. Its job is to keep machines producing.
Historically these were managed independently, often by different teams with different priorities. That's changing, and not always for comfortable reasons. Manufacturing has been the most targeted sector for cyberattacks for four consecutive years, absorbing roughly a quarter of all global ransomware incidents, and about three-quarters of those attacks originate on the IT side before reaching into OT.[4] Industry-wide, only about one in five manufacturing firms is considered "advanced" in securing a converged IT/OT environment.[5] The convergence itself isn't optional anymore — plant floors depend on IT infrastructure whether or not the organization chart treats them as connected.
Ignition Changed the Equation
One of the platforms most responsible for that convergence is Ignition, from Inductive Automation. Ignition connects PLCs, HMIs, databases, MES, ERP, and cloud platforms into a single application layer, and it has become one of the primary bridges between OT and IT on the plant floor. It's now installed in more than 140 countries and used by roughly 69% of Fortune 100 companies.[6] That scale matters here for one specific reason: Ignition, and platforms like it, only work as well as the network underneath them.
Every one of those integrations — a PLC writing production counts, an HMI polling for status, a barcode scanner uploading a batch record — is a real-time transaction riding on the plant's cabling, switches, and wireless. Industrial control literature is consistent on this point: SCADA and HMI systems are latency- and loss-sensitive by design, and degraded links introduce delayed or stale data, timeouts, and dropped sessions in exactly the systems production depends on.[7] A network that's "good enough" for email and browsing is frequently not good enough for the OT traffic riding alongside it.
Ignition bridges OT and IT — but the physical network in the middle carries every transaction in both directions, in real time.
Where OEE Fits In
Manufacturing leaders already have a metric for exactly this kind of loss: Overall Equipment Effectiveness. Industry benchmarks put "world-class" OEE at 85% or higher, built from roughly 90% availability, 95% performance, and 99%+ quality — but most manufacturing plants actually operate somewhere between 55% and 65%.[8] Availability losses — unplanned stops, in plain terms — are one of the three components of that score, and a network that can't reliably carry PLC, HMI, and MES traffic is a direct contributor to availability loss, even though it rarely shows up labeled as a "network issue" on the downtime log.
That's the connection most network conversations skip. Network engineers report on packet loss and uptime percentages. Plant managers report on OEE, scrap, and throughput. They're often describing the same underlying failure in two different languages.
Most manufacturing plants operate well below the 85%+ OEE considered world-class. Unplanned network-driven stops directly reduce the availability component of this score. Source: Fabrico / Tractian OEE benchmarks.
Not Every Plant Feels It the Same Way
None of this holds up as a blanket claim, and it's worth being honest about where it doesn't. A manufacturer running well below its maximum capacity has slack to absorb a network hiccup — the order still ships, the customer still pays, and the disruption shows up as absorbed cost (overtime, expedited freight, inventory reconciliation) rather than lost revenue. That's a legitimate distinction, not a rationalization: the "downtime equals lost revenue" assumption only holds when a plant is running close to the ceiling of what it could produce.
One manufacturer we spoke with put a number on it directly — they estimate roughly a 50% OEE hit during a network outage, and told us that at their current utilization, that hit gets absorbed rather than costing them a sale.
Two other distinctions matter just as much.
Not all equipment needs the network to run. A lot of production equipment — many printing presses and standalone packaging lines among them — has its own local, closed-loop control and doesn't depend on the plant network to physically operate. If the network drops, the machine keeps running; what's lost is the network-dependent layer on top of it — production counts, job data, MES reporting, remote diagnostics — not the output itself.
Uptime isn't the only failure mode, and it matters more in some verticals than others. Industrial engineers draw a sharp line between IT-style tolerance and OT-style tolerance: office systems typically tolerate delays of 1 to 5 milliseconds, while many operational technology processes need far tighter timing, and even a single millisecond of delay can disrupt precision robotic synchronization.[10] Automotive and robotics environments make the case starkly — factory automation, motion control, and safety-critical systems often require microsecond-level latency with essentially zero tolerance for jitter, which is exactly why standard Ethernet needed an entirely new standard, Time-Sensitive Networking, layered on top of it.[11] One vendor's real-world numbers show the mechanism plainly: under load, standard Ethernet jitter can reach 10 to 100 milliseconds, and for a servo drive synchronizing on a 250-microsecond cycle, that isn't a delay — it's a collision that trips a protective stop.[12] Printing and packaging, by contrast, rarely has that kind of tight machine-to-machine coordination, which is consistent with what our manufacturing contact told us: for his plant, it really is closer to a binary up/down question, not a latency question.
Illustrative, not to linear scale — the actual gap between IT and OT tolerance spans orders of magnitude. Standard Ethernet jitter under load (10–100 ms) can exceed an entire OT sync cycle. Source: Automation World / PatSnap Eureka.
So the honest segmentation isn't "the network matters" versus "it doesn't." It's that some verticals are mostly uptime-sensitive, and some are sensitive to the quality of a connection that never technically goes down at all — and the right way to make this case depends on knowing which one you're talking to.
Where the case doesn't disappear — it moves. Even for an uptime-sensitive plant with slack capacity, the impact doesn't vanish; it shows up as cost instead of lost revenue, and it shows up as a ceiling on how much new business the plant can take without adding a shift or a line. And that slack has a way of shrinking. Larger manufacturers tend to run closer to capacity, on tighter delivery schedules with real contractual penalties — which is exactly why the revenue case gets stronger, not weaker, as a company grows.
That's also where the strongest argument for a plant running "good enough" today actually lives: not in today's uptime, but in what's coming. Industry 4.0 investment isn't a future abstraction — the market is on pace to roughly double by 2030, and manufacturers are already committing real budget to it: one recent survey found 78% of manufacturers now put more than a fifth of their improvement budget toward smart manufacturing initiatives.[13] The gap between adopters and laggards is already measurable — one analysis found first-mover manufacturers embracing Industry 4.0 could see cash flow gains north of 100%, while non-adopters face real decline.[14] Every one of those initiatives — added sensors, MES upgrades, vision-based quality systems, predictive maintenance — adds continuous, higher-density network traffic on top of whatever's running today. A network that's fine for a mostly-standalone plant right now is a different question once that plant starts connecting the equipment it hasn't connected yet.
Validating the Numbers Against a Real Install
To ground this in something other than industry averages, we reviewed an actual nationwide cabling and access-point installation proposal for a manufacturing and warehouse facility. The scope: 15 wireless access points, 3 IDFs, 1 MDF, three fiber backbone runs, 19 Cat6 drops, roughly 4,000 feet of Cat6 and 2,000 feet of OS2 fiber — all for just under $34,600 in labor and materials.
The detail worth noticing is what wasn't included. The proposal explicitly excluded firewalls, switches, wireless hardware, licensing, support, and monitoring — the customer still had to source and own all of that separately, along with its lifecycle. That single document validated something we'd suspected: most organizations significantly underestimate the true, fully-loaded cost of owning a network, because the labor-and-cabling quote is only ever part of the bill.
A Better Comparison Than "Cheap vs. Better"
That gap is why a simple traditional-vs-better comparison undersells the decision. We now model three tiers instead:
Traditional Network
Lowest upfront cost, minimal AP density, minimal switching and redundancy, reactive support. Optimized to reduce CapEx, not to reduce downtime.
Operationally Optimized Network
A vendor-neutral engineering target built around outcomes: higher AP density, more IDFs, enterprise Cat6, fiber backbone, high availability, and real monitoring.
Meter
A fully managed implementation of that target architecture. Meter is a network-as-a-service company that designs, builds, installs, and continuously operates enterprise networks — including its own hardware, software, and support — rather than selling equipment for a customer's IT team to manage.[9]
Conceptual illustration, not measured data — the general relationship our engineering model is built on: as network investment rises across tiers, exposure to downtime-driven revenue loss falls.
Comparing those three tiers on a common, business-familiar metric — cost per square foot, five-year TCO, and estimated cost of downtime avoided — turns "we need a better network" into a number an executive can actually put in front of a board.
The Bigger Point
Downtime already costs manufacturers hundreds of thousands of dollars an hour on average. Cyberattacks increasingly cross from IT into OT because the two are converged whether or not anyone planned it that way. And the platforms — like Ignition — that make modern manufacturing possible are only as reliable as the network carrying their traffic.
The conversation shouldn't start with "how much does the network cost?" It should start with "what does an unreliable network already cost us, and what would it take to stop paying it?"
That's the case for treating the network as production infrastructure — not an IT line item.
Sources
- Aberdeen Group manufacturing downtime benchmark, corroborated by Siemens True Cost of Downtime (2024) and multiple 2025–2026 studies — info2soft.com
- L2L, 2025 Report: The Impact of Manufacturing Downtime — l2l.com
- Fluke Corporation, 2025 global manufacturing downtime survey — reliability.fluke.com
- IBM X-Force Threat Intelligence Index (2025) and related industry reporting on manufacturing cyberattack share and IT-to-OT attack paths — totalassure.com, zeronetworks.com
- Manufacturing IT/OT security readiness data — cyberpractices.org
- Inductive Automation company statistics — automation.com
- Industrial network latency/packet-loss literature on SCADA-HMI reliability — ScienceDirect, PatSnap Eureka
- OEE world-class and industry-average benchmarks — Fabrico, Tractian
- Meter company overview — Wikipedia, meter.com/about
- IT vs. OT latency tolerance in industrial networks — Automation World
- Time-Sensitive Networking requirements for motion control and robotics — PatSnap Eureka
- Standard Ethernet jitter vs. servo synchronization cycle times — maisvch.com
- Industry 4.0 market size and manufacturer investment allocation — Research and Markets, iFactoryApp / Deloitte 2025 Smart Manufacturing Survey
- McKinsey Industry 4.0 first-mover cash flow analysis — iFactoryApp
Carrier Hub is a vendor-neutral technology advisor. We evaluate a facility's network against operational outcomes — uptime, OEE, and total cost of ownership — before recommending any specific architecture or vendor, including Meter.