A connected worker is an operator whose daily work, machine status, and confirmations are shared automatically with the rest of the shop. Instead of a supervisor walking the floor to check what is happening, or an operator hunting for a paper traveler, the information is already there on a screen. This guide explains what that looks like in a small or mid-size CNC shop, why it matters, and how to set it up step by step, without a big IT project.
TL;DR:
- Start with 1 or 2 machines to keep the project small and prove it works
- Focus first on real cycle times and who is overloaded, not on tracking everything at once
- Use simple telemetry first (spindle on/off, part counts), then add more detail later if it helps
What a Connected Worker Actually Means
Think of it as the difference between a paper map and a navigation app. Both show you the route. A navigation app also shows live traffic, tells you when you have missed a turn, and adjusts as conditions change. A connected worker setup does the same thing for an operator: it shows the current state of the machine, the next task, and any alert, updated as things happen instead of once a shift.
This is different from just handing someone a tablet with a checklist. A checklist app only shows information. A connected worker system also collects information back: it logs when a machine stopped, why, and what the operator did about it. That two-way flow is what makes the data useful for planning the next day, not just recording the last one.
For a broader look at what this means for CNC shops specifically, see our article on the connected worker in CNC machining, and for the bigger picture on workforce management, see connected worker and the future of workforce management.
Why This Matters for a Small or Mid-Size CNC Shop
Most small shops cannot simply hire their way through more orders. The realistic path to more output is getting more from the team and machines already in place, by cutting the time lost to waiting: waiting for a part number, waiting for a supervisor to confirm a priority, waiting to find out a machine has been sitting idle for twenty minutes.
Two things matter most here:
Accurate cycle times. Every CNC program has a theoretical run time, but the real time on the floor is often different once you account for tool changes, first-piece checks, and small delays. Program-based estimates commonly differ from the actual runtime by 5 to 25 percent once you measure it on the floor. A connected worker system compares the two and gives you a number you can actually trust for quoting and scheduling.
Clear priorities without guesswork. When an operator knows exactly what to run next and a supervisor can see who is overloaded before it becomes a problem, fewer decisions get made on gut feeling. Shift handovers also get easier, since the system keeps a record of what was finished and what is still open, instead of relying on a note left on a machine.
What You Actually Need to Get Started
You do not need to connect everything on day one. Most shops get useful results from three simple sources of information:
| Source | What it tells you | Effort to set up |
| Basic machine signal (spindle on/off, part count) | Whether a machine is actually running, and how many parts it made | Low |
| CNC program reading (G-code) | An expected cycle time to compare against the real one | Medium |
| Operator confirmations (start, stop, reason for a pause) | Why time was lost, not just that it was lost | Low, but needs a simple screen at the machine |
Start with the first source. It is the fastest to set up and already answers the most common question on a shop floor: is this machine actually producing right now? Add program reading once you want trustworthy standard times for quoting, and add operator confirmations once you want to know the reasons behind the gaps, not just the size of them.
How to Set It Up, Step by Step
Step 1: Look at what is actually happening today. Before adding anything, spend a day or two noting where time gets lost: waiting for a part number, walking to ask a question, unclear priorities. This short exercise, sometimes called an operator workload diagnostic, tells you where a connected worker system will help most. We have a 15-minute diagnostic you can use for this.
Step 2: Pick one or two machines and set simple goals. Keep the pilot short, around 30 to 60 days, and pick a small group of operators who are willing to try it. Define what success looks like in plain terms: fewer manual interruptions, a smaller gap between planned and actual cycle time, more jobs finishing on time.
Step 3: Connect the machines and check the numbers. Start with the basic signal (spindle on/off, part count). Once that is working, compare the CNC program's expected cycle time against what the machine actually did. Do not assume the program time is correct. It rarely is, once you account for real conditions on the floor.
Step 4: Give operators a simple screen. A tablet or a small screen at the machine should show what to run next and let the operator confirm a step or flag a problem in a couple of taps. Keep it short. If it adds more clicks than it saves, operators will stop using it.
Step 5: Connect to your scheduling or ERP/MES if it makes sense. This step is optional and only worth doing once the pilot shows real value. Most shops start by syncing just the basics: when a job starts, when it finishes, and how many parts were made. Our MES-ERP integration maturity diagnostic can help you figure out if you are ready for this step or if it is better to wait.
What a Small Pilot Actually Looks Like
To make this concrete: a shop running 10 to 15 CNC machines might start by connecting just 2 of them, the ones with the most order volume or the most confusion around priorities. In the first two weeks, the goal is simply to get clean data: is the basic signal reliable, does the part count match what actually came off the machine. From week three, the shop compares real cycle times to the program's expected times and starts noticing patterns, for example a specific part number that always runs slower than planned because of a tool change step nobody accounted for. By the end of a 30 to 60 day pilot, most shops have enough evidence to decide whether to expand to more machines, and which of the two next steps (better cycle time data, or a simple operator screen) will help the most.
Common Mistakes to Avoid
Trying to track everything at once. More data is not automatically more useful. Focus on the few numbers that actually explain lost time: real cycle time, idle time, and the reasons behind manual interruptions. For more on this specific point, see our guide on reducing manual interventions.
Thinking a tablet alone solves the problem. Hardware without a link back to the schedule or the machine data only shows information, it does not close the loop. The value comes from connecting what the operator does to what the machine reports.
Skipping the operators in the design. A screen that adds clicks without saving time will get ignored. Test the interface with the people who will actually use it before rolling it out shop-wide.
Assuming you need to replace your ERP or MES. Most shops can start with basic machine telemetry and a simple screen at the machine, without touching their existing systems. You can always connect to ERP/MES later, once the basics are working.
The Bottom Line
A connected worker setup links operators, machines, and the schedule so everyone works from the same real-time picture instead of guesswork. Start small, with one or two machines and a simple screen, validate your cycle times against what actually happens on the floor, and only expand toward full ERP/MES integration once the basics are proven.
Will a connected worker system slow operators down?
Not if it is set up well. A good screen at the machine should save time by removing searching and waiting, not add extra clicks. Start with the smallest set of confirmations you actually need, for example just a start and a stop button, and watch how operators actually use it during the pilot before adding anything else. If a step gets skipped or ignored, that is a sign to simplify it rather than to add training around it.
Can this read cycle times directly from my CNC programs?
Yes. A system can read the G-code to get an expected cycle time, then compare it to what the machine actually did once it runs. The program gives you a starting estimate, but it typically misses small real-world factors like tool changes, first-piece checks, or manual adjustments. Comparing the two over a handful of runs gives you a validated number you can actually use for quoting and scheduling.
Do I need to replace my ERP or MES to start?
No. Most shops begin with basic machine signals and a simple screen at the machine, then connect to ERP or MES later once the pilot proves useful. You do not need to plan a full system replacement to get started, and in most cases the earliest wins, like knowing which machines are actually running, do not touch your existing systems at all.
How does this help with shift handovers?
The system keeps a record of what was finished, what is still open, and any issue that came up, so the next operator does not have to rely on a handwritten note or guesswork. Instead of a verbal handoff that can lose details, the incoming operator sees the exact state of each job and can pick up where the previous shift left off within a few minutes.
What should I start tracking first?
Three things matter most at the start: the real cycle time compared to the program's expected time, how much time a machine is actually idle, and the reasons behind manual interruptions. Once these three numbers are reliable and you trust them, you can add more detail, like tool life or energy use, if it turns out to be useful for your shop.
How long does a pilot typically take?
Most shops run a pilot for 30 to 60 days on 1 to 2 machines. The first couple of weeks are mostly about making sure the basic data is accurate and reliable. The remaining weeks are spent comparing real cycle times to program estimates and testing the operator screen with the team, before deciding whether to expand to more machines.
What if my shop has older CNC machines without modern connectivity?
Older controllers can usually still be connected through a simple discrete I/O signal, such as a spindle-run contact, even without built-in networking. This will not give you the same level of detail as a newer control with MTConnect or OPC-UA, but it is enough to track uptime and part counts. Our edge gateway comparison covers the hardware options for connecting older machines.
Related Articles
