A line supervisor who only sees the problem once production has already stopped is a reactive supervisor. A supervisor who sees the deviation emerging — before it stops — is a supervisor in control. The difference between the two is not experience; it is visibility. And visibility does not come from coffee-break chats or from bulletins printed on Monday. It comes from real-time data, well interpreted, within reach of whoever decides.
We have worked in dozens of Portuguese factories — textiles in Famalicão, footwear in Felgueiras, metalwork in Aveiro — and we always see the same pattern. The best plant managers are not afraid of numbers. What they fear is not having them. Because without them, all management is guesswork disguised as experience.
The invisible cost of late reaction
At a garment manufacturer in the Vale do Ave with around 120 employees, the line supervisor noticed problems when the customer called. A sewing machine went out of alignment, quality dropped, the batch fell behind, the customer complained, the production manager rushed around. All of this was "normal". Until it wasn't.
When we started looking at the raw data — cycle time per piece, recorded stoppages, defects per shift — we saw that the machine went out of alignment, on average, every 3.2 hours. No one reported it. No one recorded it. The operators simply adjusted it, lost 6-8 minutes, and carried on. Multiplied across 22 working days, 3 machines, 4 overlapping shifts: that was around 40 hours of lost production per month. In terms of customer time, that was around 8 days of potential delay.
The difference between a supervisor who reacts and one who anticipates is the time between the deviation happening and the supervisor knowing about it. Today, that time can be zero.
What changed? It was not hiring more people. It was not buying new machines. It was recording — in a structured way, in real time — what was happening. A simple terminal on the shop floor, connected to KORA Productivity, allowed the supervisor to see, on a screen, the stoppage history of her line. Not to punish — to anticipate. If the machine went out of alignment every 3.2 hours, she knew that by the third hour she needed to call the maintenance technician. Problem solved before it created a delay.
This is not a revolution. It is control. But control that works requires three things: data captured, data accessible, data interpreted at the point of decision.
What a line supervisor really needs to see
They don't need dashboards with 47 charts. They need 4-5 numbers that answer the question: "Am I on track or not?"
For a production line, those numbers are clear. Current vs. planned cycle time: if planning says 8 minutes per piece and you're running at 9.2, there's a problem. If you're at 7.8, you can tighten the planning for the next batch. Defect rate: the absolute number doesn't matter — what matters is the deviation from the target. If the target is 2% and you're at 3.8%, there's a pattern. It could be raw materials, it could be a machine adjustment, it could be operator fatigue. Stoppages and duration: how many times did it stop? For how long? Scheduled or unforeseen? A supervisor who knows this can predict when the next delay will happen. Capacity consumed: if the line has 8 hours of capacity and you've already used 7.2, you know nothing more fits today. You can warn planning.
These numbers don't need to be pretty. They need to be right and up to date. We see many factories that have very pretty dashboards, with charts that no one looks at because they arrive 24 hours late. A simple screen, with numbers from right now, is a thousand times more useful.
The problem is that most general-purpose ERPs — those that promise to "serve every sector" — cannot capture this with the granularity that a garment or footwear factory needs. Because in a generic ERP, a "production run" is a production run. In a garment factory, a production run is: 800 different SKUs, in 3 colours, in 5 sizes, with 4 different fitting layouts. The data that matters is not what the generic ERP collects.
That's why, 5 years ago, we believed a good ERP was enough. We were wrong. An ERP is the record. But real-time capture from the shop floor — what really allows the line supervisor to anticipate — is something else. It is predictive analytics applied to the next minute, not to next month.
The supervisor who anticipates is no superhero; they are someone with information
There is a belief — still very much alive in Portuguese industrial SMEs — that the best manager is the one who "senses" problems. Who knows, by intuition, when something is going to go wrong. This is true for around 5% of problems. For the other 95%, intuition is just a lack of data disguised as wisdom.
A competent line supervisor, with data in hand, will always outperform a "superhero" without information. Because data reveals patterns. And patterns enable anticipation.
We saw this at a metalworking factory in Marinha Grande. The shift supervisor was excellent — 18 years with the company, knew each machine like the back of his hand. But he was working blind. When we started recording setup times per job type, production cycles, tool consumption, we saw there was a clear correlation: whenever customer X requested a job with specification Y, there was a spike in defects 2 hours later. Why? Because the machine setup for that specification was poorly documented, and the operator did it "his own way". The supervisor, with 18 years of experience, had never seen this because he had never seen the aggregated data. He saw one stoppage here, another there, but never the pattern.
When we showed him the pattern, he was the first to say: "We need to standardise the setup." And he did. The result: defects dropped by 40%. And he didn't become any less experienced — he became more informed.
Experience without data is anecdote. Data without experience is noise. The truth lies at the intersection.
This means that implementing real-time visibility on the shop floor is not a threat to the experienced supervisor. It is a promotion. Because it frees them from looking for a needle in a haystack and lets them focus on what they do best: anticipate, decide, correct.
Why this fails in many factories (and how not to fail)
The reason why many "real-time visibility" implementations fail in Portuguese industry is not technical. It is political.
A line supervisor who sees a terminal recording every stoppage, every defect, every cycle, feels watched. Not out of paranoia — out of experience. Because they know that information can be used against them. "You can see your line had 8 stoppages yesterday." True. But the context? Did the raw material arrive defective? Was there a shortage of components? Did the previous shift leave the machine out of alignment? The raw data doesn't say.
That's why, in many factories, the line supervisor (passively) sabotages the capture. They don't record. They record late. They record wrong. And the visibility that should help becomes an instrument of pure control.
This is an implementation error, not a technology one. The technology is right. What fails is the communication. If the supervisor understands that the data is there to help them anticipate, and not to punish them, they behave differently. If they see that the factory uses the data to "catch" mistakes, rather than prevent them, they clam up.
We maintain that all data capture on the shop floor must have one clear rule: data serves to improve, not to discipline. If a line supervisor sees that a stoppage was recorded, they know they can use that information to call the maintenance technician before the next one. They do not need to fear being punished for having stopped. The difference is small in words, but enormous in practice.
Retaining talent also depends on visibility
There is a side that many managers ignore: the relationship between information and job satisfaction. In 2023, the average voluntary turnover in organisations in Portugal was 10.6%, with 52% of companies admitting difficulty in retaining talent. Source: Mercer, 2023. But there is a less obvious factor. An operator who works blind — who doesn't know whether they are meeting targets, whether their machine is well adjusted, whether the next delay is their fault or the raw material's — feels invisible. And invisibility is a silent reason to leave.
A line supervisor who can tell an operator "see? Your machine is at 7.8 minutes per piece, the target is 8, you're 2.5% above" — is giving them feedback. They are showing them that their work is measured, that it is seen, that it matters. This is not manipulation. It is structured recognition. And structured recognition reduces turnover.
Anticipating is not guessing: it is measuring and reacting
A supervisor who anticipates deviations does three things, in this order.
First, they measure. They capture data on what is happening. Not at the end of the day — now. Not on paper — in a system. Not in aggregate — in enough detail to see patterns. This requires a terminal on the shop floor, connected to a production capture system that doesn't slow the operator down. If the terminal takes 30 seconds to record a stoppage, the operator won't use it. If it takes 3 seconds, they will.
Then, they compare. They match what is happening against what should be happening. Cycle time vs. planned. Defect rate vs. target. Capacity consumed vs. available. This requires a clear plan — if you don't know what the target is, you can't tell whether you're above or below it. Many factories have data but no targets. Or they have targets that no one knows. This is as useless as having no data at all.
Finally, they act. They don't wait for the end of the shift, for the Monday meeting, or for the monthly report. They act in the next minute. They call the technician, readjust the machine, switch operator, warn planning. Because 10 minutes of delay now become 2 hours of delay if they aren't corrected. This requires the supervisor to have the authority to decide — and the factory to have the resources to respond. If the supervisor sees a deviation but the maintenance technician is on another line, the system fails.
This is only possible if the supervisor has access to real-time data. They don't need a complex system. They need a terminal on the shop floor, connected to a vertical ERP that understands the granularity of their sector, and a culture that says: "Data is there to improve, not to punish."
Real-time visibility is not a luxury for large factories. It is a necessity for a factory that wants to compete. Because the factory that sees the deviation in 30 seconds will always beat the one that sees it in 30 minutes. And that difference, multiplied across 22 working days, is the difference between being on time and being late.
Frequently asked questions
What is real-time performance management on the shop floor?
It is the structured capture of operational data — cycle time, stoppages, defects, capacity — at the moment it happens, made available to the line supervisor on a simple screen. It allows deviations to be anticipated before production stops, rather than reacting after the customer complains. It is not about having lots of charts; it is about having the right numbers, now.
What is the difference between a reactive supervisor and one who anticipates?
A reactive supervisor sees the problem once production has already stopped. One who anticipates sees the deviation emerging beforehand. The difference is not experience — it is visibility. A supervisor with real-time data can predict that a machine will go out of alignment by the third hour and call the technician preventively, avoiding delay.
How many indicators should a production line dashboard have?
It doesn't need 47 charts. Just 4-5 numbers are enough: current vs. planned cycle time, defect rate vs. target, stoppages (how many and for how long), and capacity consumed. These indicators answer the essential question: "Am I on track or not?" Simple, correct and up-to-date numbers are worth more than pretty dashboards with data from 24 hours ago.
Why can't generic ERPs capture data with the necessary granularity?
A generic ERP sees "a production run" as a single unit. In a garment factory, a production run is 800 SKUs in 3 colours, 5 sizes, 4 different fitting layouts. The data that really matters — time per piece, defects per shift, stoppages per machine — is not captured with the precision that a specialised factory needs. An ERP is a record; real-time shop-floor analysis is something else.
Does an experienced supervisor feel threatened by the implementation of real-time data?
No. Real-time data is not a threat — it is a promotion. It frees the supervisor from looking for a needle in a haystack and lets them focus on what they do best: anticipate, decide, correct. Experience without data is anecdote; data without experience is noise. The truth lies at the intersection of the two.
How do you calculate the real cost of not having real-time visibility?
At a garment manufacturer with 120 employees, a machine went out of alignment every 3.2 hours — losing 6-8 minutes per adjustment — but no one recorded it. Multiplied across 22 days, 3 machines, 4 shifts: that was 40 hours lost per month, equivalent to 8 days of potential delay. The invisible cost is the sum of small inefficiencies that no one sees in isolation.
What is the difference between generic predictive analytics and the kind that works on the shop floor?
Generic predictive analytics forecasts next month's trends. The kind that works on the shop floor forecasts the next minute's deviation. If a machine goes out of alignment every 3.2 hours, the supervisor knows that by the third hour action is needed. This is operational anticipation — not strategic — and it requires data captured, accessible and interpreted at the point of decision.