JABS Sokółka produces 1,000 more windows a month, without buying a single new machine
JABS Sokółka is a wooden window manufacturer based in Poland, with over 50 years of production history. It is the largest plant in JABS Group, part of Inwido, Europe's biggest window and door group, and produces around 70,000 made-to-measure windows a year for customers across Scandinavia and Europe.
The factory runs six different window platforms on the same floor, with every unit built to a customer order and no two orders identical. Roughly half of its 50 machines are more than 20 years old. Until recently, production tracking was limited to department-level output figures and pen-and-paper counts. Output, downtime and changeovers per machine were not measured at all.
In 2024, JABS Sokółka connected its first set of most important machines to GlobalReader, starting with the line producing 70% of its capacity.
Two years later, and their daily output is up from 280 to 340 windows, efficiency has grown 20-25%, and investment decisions are now made based on real machine data instead of estimates.
In this case study, Maciej Czajka, Production Manager at JABS Sokółka, tells us exactly what it takes to run a high-mix factory where every window is custom: how they measured performance before, what finally pushed them to look for a monitoring solution, and how they now produce 1,000 more windows a month with GlobalReader.
Company profile:
What they produce: Premium made-to-measure wooden and wood-aluminium windows (six different window platforms, every unit built to a customer order)
Company size: 350 employees (300 in production), they are part of Inwido (Europe's largest window and door group)
Customers: Sold direct-to-consumer online under JABS Group brands, mainly Scandinavia (Denmark, Norway, Sweden), Germany, the Netherlands, Austria and Poland
Production volume: 70,000 windows produced annually
Shifts: 2 shifts across three departments (machinery, paint lines, assembly/glazing), plus Saturdays in peak season
Machines: 50 total, of which 25 are older than 20 years
Connected to GlobalReader: 12 machines (started with 6 machines covering 70% of production capacity, plus 3 paint lines, expanding step by step)
Most important KPIs for JABS: Output, availability (interruptions and their reasons), performance, OEE per machine
How they use GlobalReader: Real-time output and downtime monitoring, interruption reason tracking, OEE, custom energy monitoring (amperage per machine), operator panels on phone/tablet/PC, multi-factory access for cross-plant benchmarking, Polish-language interface
Systems landscape: GlobalReader runs alongside the plant's existing maintenance software (GlobalReader's maintenance module and scrap tracking are on the future roadmap)
Before GlobalReader: A factory running on pen, paper and gut feeling
Before GlobalReader, JABS Sokółka tracked production the way it had for decades. Each department reported its total output, but nobody could say what any individual machine produced, how often it stopped, or why.
“If a machine had maintenance for two hours, it was just a short mention at the next morning meeting.”
The gaps showed up everywhere in daily operations. If management needed machine-level data, the only option was manual: an operator wrote his piece counts on paper, and someone in the office then had to type and process the data by hand. This meant the information was often late, incomplete, and too expensive to collect regularly. Machine downtime surfaced a day later, as a brief verbal update at the morning meeting, with no record of shorter stops, micro-stops, or their causes.
“We used pen and paper. The operator would write: today I made 55 pieces. Then you needed somebody in the office to work with that data. That was the most difficult part.”
This created two concrete problems.
First, output swings were unexplainable: when Monday produced less than Tuesday, nobody could say why, so nobody could fix it.
“Before installing GlobalReader, we didn't even know that we had a problem. We didn't have information why Monday's output was worse than Tuesday's, or the other way around.”
Second, unmeasured time quietly disappeared on the floor. Extended breaks and small stops that no one could see.
The result: a 350-person factory making 70,000 windows a year, with no way to explain why one day was good and the next was bad.
The biggest bottleneck: Knowing the problem, but having zero proof
The most expensive consequence for JABS was decision-making without evidence.
“We knew that machine was our bottleneck. But we didn't have any numbers.”
The team knew one machine was the plant's bottleneck, but with no numbers, they couldn't prove it, size it, or build an investment case around it.
Planning ran on Excel assumptions that turned out to be wrong: the spreadsheet said 45 changeovers a day on a key machine but the reality was between 60-70. And on the floor, unmeasured time quietly disappeared: micro-stops and stretched breaks that nobody could see, on top of operators who had been flagging problems for years with no data to back them up.
Why did JABS Sokółka choose GlobalReader?
The plant had spent three years working hard on process improvement but without machine-level measurement, nobody could see whether changes worked, where time was lost, or why output swung from day to day.
It was clear to the production team that in order to truly improve the factory’s efficiency, they needed to start measuring the processes.
“The biggest reason, we started looking for a factory monitoring software, was to start measuring the process, so we could improve the process.”
The search became a group decision. JABS’s Estonian sister plant was already running GlobalReader, so when Sokółka began evaluating tools, they compared several suppliers side by side. What tipped the decision was usability: software that operators, foremen and managers could actually learn and run themselves, on a phone, tablet or PC, without months of consultants.
“We had meetings with different suppliers, we compared prices and options. What made the difference is that GlobalReader is very easy to use, and very easy to learn. And GlobalReader has a Polish language option! For operators, that’s very important.”
JABS installed the GlobalReader hardware themselves across the full mix of machinery, from modern machines more than 20 years old ones, using GlobalReader’s instructions.
On newer machines the sensors weren't even needed: the signal is taken directly from the machine's electrical cabinet. On the older ones, the team mounted the additional sensors themselves. The entire rollout, from hardware to trained operators, was managed remotely with GlobalReader's onboarding team.
“We installed everything ourselves. It’s quite easy if you have somebody who knows how the machine works.”
With GlobalReader: JABS runs on numbers, from the shop floor to the boardroom in Denmark
Today the day-to-day running of the factory looks fundamentally different.
“Finally, we have numbers. We have arguments in our pockets.”
Where machine performance used to live on paper and day-old verbal updates at the morning meeting, supervisors and foremen now watch live output and interruption data daily and report to the production manager on what stopped, and why.
Problems get corrected within the same working day. If the first shift falls behind its numbers, the team knows immediately and the second shift can make up the difference that same day.
“Since we installed Global Reader, people have been working more efficiently on their own, consistently. I believe it's a psychological factor. Since someone or something is monitoring the machine and the operator is responsible for marking downtime, they themselves strive to eliminate it.”
Repetitive interruptions get a five-why analysis (together with the operators who mark them) and the fixes go back into the line. Operators are now part of the improvement loop, because every problem they flag is recorded, visible and impossible to dismiss.
“We can now go to our boss and say: we already have 95% usage on this machine. It’s impossible to make more on it. If we want to produce more, we need to invest. Not in two machines, but only in that specific one.”
The data travels upward, too. Every month, the production manager builds a report from GlobalReader numbers for group management in Denmark, and the figures drive joint decisions on where to improve next. The same data feeds beyond production: the sales team now uses real machine-time data to calculate product costs accurately, as the plant finishes a value-added mapping project built on GlobalReader measurements.
“Every month I make a report for our headquarters in Denmark based on GlobalReader numbers, and together we discuss what to implement and which direction to go.”
And GlobalReader data catches what humans can’t. In early 2026, a supervisor noticed daily paint line numbers drifting down. The team first suspected the GlobalReader numbers were wrong, but they weren’t. It turned out that the paint line belt had slowed from 3.8 to 3.5 metres per minute, a manual setting on an old line with no measurement of its own, silently costing 50-100 traverses a day. Without GlobalReader, nobody would ever have caught it.
“First we thought something was wrong with GlobalReader. But GlobalReader was okay. The belt speed had dropped from 3.8 to 3.5 metres per minute. That’s 50 to 100 traverses a day!”
For deeper process analysis, the team developed its own method: on one screen they play the previous day's camera footage from a workstation, and on the other they open the same period's GlobalReader interruption diagram. Comparing the two side by side shows exactly where time is lost in the process (a slow material handoff, a missing trolley, a changeover that takes longer than it should) and where the workstation needs better tools or a better layout to support the operator.
“Operators want to talk. Now we see the interruptions they mark, and we ask: what should we do about this? It’s very good for both sides.”
Lastly, the clearest example of the new workflow is the paint line cleaning routine (changing filters and washing the line at the start and end of every day). All of this used to take one hour in the morning and another in the evening. Because GlobalReader recorded this as a large, repeating interruption, the team analyzed it with the operators who run the line and agreed on a better way to do the work. Today the same cleaning takes 30 minutes in the morning and 45 in the evening. That means 45 minutes of production capacity recovered every single day.
JABS’ results with GlobalReader: From 280 to 340 windows produced per day on the same machines
Two years in, the numbers speak for themselves.
“GlobalReader doesn’t give you the solution to your problems. It gives you the information to understand that you actually have a problem. Before, you didn’t even know.”
As the production director puts it, GlobalReader isn’t the only driver, the team’s process improvement work matters, but it’s the system that made that work measurable, provable and repeatable:
✅ +12,000 additional windows produced per year: daily output grew from 280 to 340 windows, a 21% increase with no new production machines, no extra floor space or new hires
✅ +20-25% efficiency in two years: output per working hour, not just total volume
✅ 38% less daily cleaning downtime on the paint line: from 2 hours per day (1h start + 1h end) to 75 minutes, recovering over 150 hours of paint line capacity a year
✅ 50-100 traverses per day recovered by catching a silent belt-speed drift (3.8 to 3.5 m/min) that no manual system would have spotted
✅ One confident data-backed machine purchase: with the bottleneck machine documented at 95% utilization, the investment case went to group management with proof, targeting exactly the constraint instead of guessing
But on top of these numbers come the quiet, but equally important, savings: machines producing instead of idling on standby, ventilation and electricity spent on output rather than empty time, paint no longer sprayed on empty trolleys, and a lower CO₂ footprint per window.
““Higher efficiency means lower cost of production in total. We’re producing a lot more in the same amount of time now, while saving on electricity. The ventilation isn’t running for nothing and the machines are not on standby but working.””
