Blog » Digitalisation

Your first cobot on the shop floor: how to choose the operation and calculate the payback

Digitalisation · 5 min read · 05.11.2026

In a hall we know well, a cobot stands in the corner, wrapped in stretch film. They bought it three years ago after a trade fair. They ran it twice. Today it holds drawings. The arm is fine, the software too — it was the choice of operation that failed. The cycle changed every other week, three variants were added to the product mix, and nobody ran a time study before signing the purchase order. The technology wasn't the problem. The problem was that nobody could say what the cobot was actually supposed to do.

This case is not unusual. Small and medium-sized manufacturers lag behind the large ones in robotization: automated lines are something industry leaders can afford, while smaller companies wait until their labor costs keep rising and until they can no longer find the last machine operator on the labor market. At Foreast, we see something different in our projects. For your first robotization, you don't need a smart factory — you need one well-chosen cobot on a repetitive operation that pays back in 12 to 24 months. But only if you choose it based on data from the shop floor, not on a demo at a trade fair.

Where a cobot pays off: five criteria for choosing the operation

Every successful first deployment has a similar signature: a stable operation, high turnover in the position, and a cycle that can't be messed up. Go through the five criteria before your first contact with a supplier.

The first is the cycle. Identical motions at a takt of 15 to 120 seconds. If the operation changes with every order, automation can't handle it without expensive peripherals. The second is staffing: a cobot running on three shifts replaces three full-time positions, while an operation staffed on just one shift takes five years or more to pay back. The third is the product mix. The product runs for at least 12 to 18 months, with orders confirmed long-term. A cobot for prototypes is a toy, not an investment.

Fourth: variability. At most two or three part variants, reprogramming within 15 minutes, parts arriving in a stacker rather than mixed "however they come." And fifth, last: no sensory inspection. A cobot can't see a scratch or feel a burr. Where the operator judges quality by eye, that part of the process stays with a human.

A cobot won't improve a bad process — it will only repeat it faster. That's why the choice of operation matters more than the brand of the arm.

You can calculate the payback in a single morning

Cobot suppliers receive inquiries like "we might be interested in automation, what do you recommend?" every day. Without numbers, they can't respond. Yet all six data points are right there in your workshop, and collecting them takes one morning: the hourly rate for the position including payroll contributions; the number of shifts and hours during which the operation is staffed; the cycle time or motions per hour; the reject rate; turnover and the average time to train a new hire; and the weight, dimensions, and number of part variants.

So what happens when you actually send the numbers? A machining company with 80 employees, tending a machining center on two shifts. The complete cell — cobot, gripper, safety sensors, integrated deployment and training — cost €48,000. The cobot freed up the equivalent of 1.2 workers, that is 2,040 hours per year × €14/h = €28,560.

Consistent part loading cut the reject rate from 2.4% to 0.7%, which at that volume meant another €3,600 per year. That adds up to €32,160 in savings per year and a payback of roughly 18 months. And the 25% annual turnover in this position isn't even in the figures. That's the safety margin.

And what if the calculation comes out at four years or more? Then a cobot is not the right fit for this operation. Find another one, or stabilize the cycle and material flow first — sometimes repositioning a few work stops is enough to open up more candidates. A cobot bought against the numbers will end up exactly like the one in stretch film.

A 90-day pilot: trial operation with an exit clause

The pilot has three phases. In the first two weeks, you do the measuring: a time study across three shifts, video of the entire cycle, micro-stops, scrap. Weeks three to six belong to the supplier — cell design, a quote with a guaranteed cycle time and number of interventions, a contract with an exit clause and payment tied to acceptance. Weeks seven through twelve are installation and parallel operation: the cobot runs, the operator stands by, and every intervention is logged.

Parallel operation isn't a delay. It's the only way to get real numbers on interventions without putting customer deliveries at risk.

Involve four people from day one. The foreman weighs in on the shift schedule and material flow. Maintenance takes over the power supply, compressed air, network, and service plan. The health and safety officer assesses the safety of the cell before installation, not after it. The operator gets trained on reprogramming the variants and becomes the cobot's owner.

Measure four things: motions per shift against plan, the number of interventions (the target is fewer than two per shift), the reject rate, and cell uptime. Set your stop condition in advance. If after six weeks of parallel operation you exceed five interventions per shift, the cycle is more than 10% longer than guaranteed, or the supplier is slipping on milestones, activate the exit clause. Better to send the cobot back and lose a month than to keep a drawing holder for three years.

Key takeaways

  • The operation must have a stable cycle of 15–120 seconds and be staffed on at least two shifts — otherwise the cobot won't pay back even in five years.
  • Before the first contact with a supplier, collect six numbers: the hourly rate including payroll contributions, shift staffing, motions, rejected parts, turnover, and part weight.
  • Budget €45,000 to €55,000 for a complete cell; with savings of around €30,000 per year, the payback is roughly 18 months.
  • The pilot contract must include an exit clause and payment tied to the guaranteed cycle time and the number of interventions per shift.

Your task for Monday: pick three candidate operations and spend two weeks collecting data on them — cycle time, downtime, scrap, shift staffing, turnover. Then sit down for 90 minutes with the foreman and controlling and choose one to send to the supplier with complete numbers. If you'd like to run the selection and pilot preparation with someone who has done this before, take a look at our services.

Your first cobot doesn't have to be a step toward a smart factory. It's enough that it pays back within 18 months and covers the shifts few people want to work these days.



« Previous article  |  Next article »

« All articles