The number one objection from a Brazilian industrial manager when the topic is Industry 4.0 is always the same. "My machines are from the 90s, this won't work for me." It's a legitimate fear, but built on a wrong premise. The idea that industrial IoT requires a new factory, new machines, and millions in investment is among the main myths of the sector — and McKinsey itself, in its report on the six myths of IIoT, is direct: most of the value of industrial IoT comes precisely from the factories that already exist.
The word that describes this in the literature is brownfield: installed plants, with legacy equipment, consolidated processes, and decades of operation. It's in this scenario that industrial IoT delivers the greatest return, because the marginal gain of connecting an old machine is far higher than buying a new, already-connected one. This post is about how that works in practice, across three levels of depth.
Level 1: the PLC you already have is your IoT
The good news few people tell is that most factories already have the equipment they need to start. The PLC that controls the machine, installed 10 or 20 years ago, already generates production data in real time: machine cycle, operation status, part count, alarms, readings from sensors connected to it. The data exists. The problem is that it stays locked inside the PLC, never reaching any system that can actually use it.
What you do at this level is simple: you install an industrial IoT gateway that talks to the PLC via standardized protocols (Modbus TCP/RTU, Profinet, Ethernet-IP, OPC-UA) and sends that data to a monitoring platform. You don't touch the PLC logic. You don't stop the machine. You don't replace anything. In a single morning of installation, the machine starts being monitored in real time.
These standardized protocols today cover more than 70% of existing industrial assets. In most factories, connecting isn't about replacing a machine — it's about reading what it already says.
Level 2: machines without a PLC, external sensors
But what about the 1980s press that has no PLC at all? Or that machine with a closed PLC and no communication? This is where the second level comes in: external sensing. Non-invasive sensors for electrical current, temperature, position, or produced-unit counting can be installed without touching the original equipment.
Installation is typically quick: a current sensor clamped to the motor cable already tells you whether the machine is running or stopped, for how long, and reveals operating patterns that anticipate failures. A vibration sensor glued to the housing detects bearing wear weeks before the equipment breaks. The cost per machine is usually a fraction of the cost of a single unplanned stop.
Level 3: from raw data to actionable information
Collecting data is the easy part. Turning data into information that changes a decision is where the real work lies.
A sensor sending 60 readings per minute over one shift generates 28,800 data points per machine. Multiplied across dozens or hundreds of machines, it becomes a volume no generic dashboard can handle on its own.
That's why an industrial IoT platform doesn't stop at receiving the data. It structures the data into business metrics (OEE, availability, performance, quality), detects stops automatically, classifies their reasons, generates intelligent alerts, and exposes open APIs for integration with ERP, production planning, and maintenance. Without that layer, IoT becomes a collection of pretty numbers with no practical use.
Where to start
The most common mistake in industrial IoT projects is starting with the technology, not the problem. Buying 200 sensors "to see what comes of it" is the fastest way to spend money with no results. The approach that works is the opposite: pick a concrete problem (a stop on the main press, scrap on the extruder, corrective maintenance on the critical pump), instrument exactly what needs to be measured to understand that problem, and expand from what you learned.
McKinsey reinforces this point: companies that try to scale IoT before proving value on a specific use case rarely manage to replicate results across plants. Those that start small, demonstrate return, and expand in waves manage to transform the entire operation within a few years.
The factory you have is the factory you connect
Industrial IoT isn't about a new factory. It's about extracting value from the factory that exists. The PLC that's been there for 15 years, the press that still produces, the motor nobody wants to replace — all of it can become a source of useful data in a short timeframe and with predictable investment. The question to ask isn't "is my factory modern enough for IoT?". It's another one: what is the cost for your operation to keep making decisions without knowing what each of those machines is doing right now?
