You’re short two operators on second shift and the req has been open for months. Someone suggests automating a station, and suddenly you’re staring at a dozen candidate jobs with no clear way to pick the right one. Pick wrong and you spend a budget on a cell that fights your process every day. Pick right and you free up people for the work that actually needs judgment.
The plants that get automation right don’t start with the flashiest job on the floor. They start with the job that’s easiest to automate well, fix what’s broken in the process first, and build credibility with that first win before they touch anything harder. This guide is about how to find that job, get it ready, and sequence what comes after it.
Start with dull, dirty, dangerous, and repeatable
The best first automation cell rarely comes from a whiteboard session about innovation. It comes from watching your floor and asking which jobs nobody wants to do, and why.
Dull jobs are the ones where an operator does the exact same motion hundreds or thousands of times a shift. Fatigue sets in, attention drifts, and quality suffers late in the shift even with a good operator. A robot doesn’t get tired at hour seven.
Dirty jobs put people around debris, coolant, dust, or fumes. These are often the hardest to staff and the first ones people quit over. Automating them isn’t just about output, it’s about not having an open req every quarter.
Dangerous jobs involve pinch points, presses, hot parts, or heavy lifting. These carry injury risk regardless of how careful your team is. Removing a person from that station is a safety win before it’s a productivity win.
Repeatable jobs have a consistent, predictable process every cycle. This one matters most for a first cell, because repeatability is what makes automation reliable. A job that changes in small ways every time is a much harder automation target, even if it’s dull, dirty, or dangerous.
Look for a job that hits at least two or three of these. That overlap is usually where your first cell should live.
Run a part-readiness audit before you pick a cell
Before you commit to a candidate job, walk it the way a robot would see it, not the way a person sees it. A skilled operator adapts to a part that’s slightly off or dirty from the last operation without even noticing. A robot doesn’t adapt. It does exactly what it was taught, every time, on the part it expects. A part-readiness audit tells you if the part you’re sending it matches that expectation.
Presentation. How does the part actually arrive at the station today? If it comes in a neat, indexed tray or fixture, that’s a strong sign. If it arrives in a jumbled tote with random orientation, you’re either adding vision and more engineering, or you’re changing how the part gets there before automating the pick.
Tolerance. Pull real parts from the line, not the print. Measure a sample against the tolerance your process actually needs, not the tolerance the drawing says it should hit. If the real parts vary more than the drawing implies, that gap is exactly what will trip up a robot that was taught to expect the print.
Variation. Watch ten cycles of the manual process back to back. Does the same operator do it the same way every time? Do two different operators get the same result? If the answer is no on either count, the process has more variation than it looks like from the floor, and that variation needs to shrink before a robot can repeat it reliably.
Upstream stability. A part that comes out of an unstable upstream process, one with tooling wear, material lot variation, or inconsistent machine setup, will never present consistently no matter how good the automation is downstream. Trace the part back a step or two before you commit to automating where it lands.
A job that scores well on all four is close to automation-ready today. A job that scores poorly on two or more isn’t a bad candidate forever, it’s a candidate that needs process work first.
Fix the manual process before you automate it
It’s tempting to automate a messy process exactly as it exists today, on the theory that the robot will smooth it out. It won’t. Automation makes a stable process faster and more consistent. It makes an unstable process fail the same way, just faster and with less ability to adapt on the fly.
Before you automate a station, look hard at three things. Fixturing should locate the part the same way every single time, with no operator judgment involved in getting it seated right. If your current fixturing depends on a skilled hand to seat the part correctly, that’s a fixture problem to solve before it’s an automation problem, and it’s often worth a conversation with whoever builds your jigs and fixtures before you spec the cell.
Part quality coming into the station needs to be consistent enough that the robot isn’t compensating for defects a person would catch and set aside without thinking. If your reject rate at this station today depends on an operator’s eye, quantify it before automating.
Standard work means the manual process has one clearly defined way of doing the job, written down, not three different ways depending on who’s on shift. If you can’t write the current process down in a dozen clear steps, you don’t have a process to automate yet, you have a habit that lives in a few people’s hands. Standardize it manually first. It’s cheaper to fix on paper than to discover mid-build that the “one job” was actually three jobs wearing the same name.
Why the highest-tech cell is rarely the best first cell
It’s tempting to automate the most impressive process on your floor first, the one that would make the best photo for your website. Resist that instinct. The most technically advanced job is usually the one with the most variables: tight tolerances, unpredictable part presentation, or a process that depends on operator feel and judgment. Those are the hardest things to automate well, and a rough first attempt on your hardest job will sour your whole team on automation before it gets a fair shot.
Your first cell should be a job you can automate with a high degree of confidence, using proven technology, so it works well from day one. Save the harder, more technical processes for your second or third project, once your team has hands-on experience running a cell and trusts the technology.
Sequencing a multi-year automation roadmap
Treat your first cell as the start of a roadmap, not a one-off project. The order you tackle jobs in matters as much as which jobs you pick.
Phase one: the boring cell. Pick the job that scores best on dull, dirty, dangerous, repeatable, and part-readiness. Get it running well and build the internal muscle for running a cell day to day. This is the win that earns you credibility to propose the next one.
Phase two: adjacent jobs. Once the first cell is proven, look at jobs that are similar in kind but slightly harder, maybe a bit more part variation or a second product variant. Your team now has real experience to draw on instead of guessing.
Phase three: the harder cases. With two cells under your belt, tackle the jobs that needed vision systems, more complex end effectors, or tighter integration with upstream and downstream equipment. By now you know your plant’s failure modes well enough to spec these correctly instead of learning it the hard way.
Trying to leapfrog straight to the hardest job because it has the biggest theoretical payoff is how plants end up with a project that never quite works right and a team that doesn’t trust the next one either.
What makes a job robot-friendly
A few more structural things separate a job that’s genuinely robot-friendly from one that will keep fighting you. Some parts grip easily, others need custom end of arm tooling built for their shape, weight, or surface finish, so a simple, consistent grip point beats a part that needs a multi-part gripper for several SKUs. And the job needs enough volume to justify the engineering time behind it. A low-volume job with high variability might be better served by a smarter fixture than a full robotic cell, at least for now.
The hidden costs people forget
The robot arm itself is rarely the surprise in an automation project. The surprises live in everything around it.
- End of arm tooling has to be designed for your specific part, and it often needs redesign as your product mix changes
- Guarding and safety systems around the cell add cost and floor space early estimates leave out
- Part infeed and outfeed, including conveyors, bowl feeders, or fixturing, can cost as much as the robot itself if your process doesn’t already present parts in a robot-ready way
- Integration with existing PLCs and your plant network takes engineering time that’s easy to underestimate
- Training and changeover procedures, so the cell doesn’t become a black box only one person understands
Budget for these from the start. A quote that only covers the robot and not what surrounds it isn’t a complete picture of what the cell will actually cost to get running. If the job you’re evaluating is closer to a fully custom build than a robotic pick-and-place, it’s worth comparing against what actually drives custom machine cost before you settle on an approach.
The people side: what happens to your operators
The most overlooked part of a first automation project isn’t the robot, it’s the person who used to run that station. Handled well, automation turns an operator into a cell tender, someone who oversees the cell, handles exceptions, restocks materials, and catches problems before they become downtime. Handled poorly, it looks like a threat, and you lose the cooperation you need to make the project succeed.
Involve your operators early, not after the cell is installed. The person who’s run that station for years usually knows more about the part’s real-world quirks than anyone in the room, and that knowledge should shape the design, not get discovered the hard way during commissioning.
Plan real training, not a half-day walkthrough. Operators need to understand normal cycle behavior well enough to spot abnormal behavior early, know basic troubleshooting steps, and understand how to safely intervene when something jams or faults out.
Decide who owns the robot after handoff before handoff happens. In most plants that’s a blend: maintenance handles mechanical and electrical upkeep, a trained cell tender handles day-to-day operation, and someone specific is the point of contact for anything beyond that. If ownership is still undefined on the day the integrator leaves, the cell becomes everyone’s job, which means it’s really no one’s job.
How to judge readiness
Before you commit to a first cell, ask your team a few honest questions. Is the process stable enough that two operators doing the same job get the same result? Is part presentation something you can control, or does it depend on how a person happens to place things? Does the job have enough volume to justify the engineering time? Have you fixed the fixturing, quality, and standard work issues that would follow the job into the automated cell?
If you can’t answer these clearly, that’s not a reason to abandon automation. It’s a signal the job needs process work first, or that a different job on your floor is a better starting point.
Darioo Industrial works through exactly this kind of readiness assessment with plants before recommending a first cell, because the wrong first project can set automation back years on a floor that needed it most.
Common questions
How do I know if a job is too complex for a first automation project? If the job depends heavily on operator judgment, has parts that vary noticeably cycle to cycle, or needs a custom gripper for several SKUs, it’s likely too complex for a first cell. Save it for after your team has a proven cell running.
Should I automate a low-volume job if it’s dangerous? Safety concerns can justify automating a lower-volume job sooner than volume alone would suggest, but weigh it against engineering cost. Sometimes a fixture redesign removes the danger without needing a full robotic cell. Get a straight answer on both options before deciding.
What if my part quality is inconsistent right now? Fix what you can upstream before automating. Inconsistent part quality that a person currently catches and corrects will pass straight through a robot unless you build in inspection to catch it, which adds cost and complexity. It’s often cheaper to tighten the upstream process first.
Who should be involved in picking the first automation job? Include the operators who run the candidate stations today, your maintenance lead, and whoever owns the budget. Operators catch part and process issues that don’t show up on paper. Maintenance flags support and reliability concerns early. Skipping either group is a common reason first projects underperform.
The takeaway
Your first automation cell should be dull, dirty, dangerous, or repeatable, not necessarily the most advanced, and it should pass a real part-readiness audit before you commit budget to it. Fix fixturing, part quality, and standard work in the manual process first, plan for the people side as carefully as the hardware, and sequence your roadmap so each project builds on what the last one taught your team. Get that first cell right and you build the internal trust and know-how to take on harder automation projects next.