August 2026 | volume 80
Today, Japanese manufacturing is regarded as a model of quality, and kaizen looks as though it has always existed in Japan—like tea, bamboo and Zen gardens. This is not true. As recently as the 1950s, the words “Made in Japan” were synonymous in the West with shoddy goods: unreliable, impermanent, technically outdated products of terrible quality. They were bought because they were cheap and thrown away when they broke—and they broke quickly.
W-MOSZCZYNSKI-PS-8-2026Paradoxically, it was Americans who brought change. The American statistician William Edwards Deming came to Japan, which had been ruined by the war. He did not bring money or technology. He brought an idea: improvement must be continuous, measurable and small. The Japanese did not merely listen to him—they absorbed this idea, combined it with their own culture of patience and precision, and turned it into a way of running factories. Twenty years later, the Japanese economy was one of the largest in the world, and Toyota—the company that applied kaizen most systematically—had become one of the most effective manufacturers in industrial history. Not thanks to a single revolution, but thanks to thousands of small improvements introduced every day.
For the food industry, this is a doubly important lesson. Firstly, because a reputation for quality—as the Japanese example shows—can be rebuilt, but only by the method of small steps. Secondly, because in a food plant, where a raw material’s shelf life is measured in hours and margins in single percentages, small losses accumulate faster than anywhere else.
Change is not an event; it is a direction
The word kaizen consists of two characters: kai—change, zen—towards the good. Together: continuous improvement. It sounds banal, but the essence lies not in the definition, but in the way in which this improvement occurs.
Most plants think of change as an event: a new line, a new ERP system, a large optimization project with external consultants, “from January we will do everything differently”. Kaizen proposes the opposite: change is a direction, not an event. One small step today, another tomorrow, another the day after tomorrow. An endless series of small improvements that accumulate into a result which, viewed from the outside, looks impossible.
Mathematics is merciless in both directions. An improvement of 1% per day produces a result 37 times better after a year (1.01365 ≈ 37). A deterioration of 1% per day reduces the result to 30% of the initial value. In plant practice, no one improves by one per cent a day—but a filling line on which one small cause of a micro-stoppage is eliminated every week works completely differently after a year from a line on which “somehow it kept going” every week.
Why do large change programmes so often fall apart? Not because of the crew’s laziness. A large change provokes resistance—from people, procedures and habits—and requires resources that production under deadline pressure never has. A small step does not provoke resistance. No one protests against moving a pallet of seals three metres closer to the workstation. After a year of such moves, the production hall looks different.
Genchi gembutsu: go to the production hall, not to the report
Before the first step is taken, however, it is necessary to know where one is starting from. Japanese industry follows the principle of genchi gembutsu—literally “real place, real thing”. Before you solve a problem, go to where it arises. Do not read reports or listen to second-hand opinions. Look with your own eyes.
It sounds excellent; in practice, few people do it. A production manager sees a decline in a packaging machine’s efficiency in an OEE report and orders a technical audit. Meanwhile, an hour spent by the machine would show that on every shift the operators lose twenty minutes looking for the correct film because the warehouse issues it in a random order. A logistics director analyses delays in deliveries to a retail chain in Excel and suspects the drivers, while two hours of watching footage from the plant’s cameras would reveal that the vehicles stand for forty minutes at the ramp of the company’s own warehouse, waiting for documents.
Kaizen says: before making a diagnosis, observe. Honestly, without judging, for a week. Record what is really happening on the line, at the ramp, in the cold store. Only observation determines the first step—not the general theory “efficiency must be improved”, which generally means nothing and leads to nothing, but a specific point: “film on pallets must be sorted according to the production schedule”.
Muda: find what is leaking
A key kaizen concept is muda—waste. Everything that consumes time, raw material, energy and money but does not add value for which the customer pays. The classical Japanese school distinguishes seven types of waste, and each one has its own painful, specific face in the food industry.
Overproduction is the most expensive waste in food production because the product has a shelf life. A retail chain will no longer accept yoghurts produced “for stock” if they will not leave the warehouse before two thirds of their shelf life has passed—the options left are a price reduction, disposal or a food bank.
Waiting means stoppages: the line stands because the laboratory has not yet released a batch of raw material; it stands because changeover from strawberry to peach flavour takes ninety minutes instead of thirty; it stands because the planner placed two orders requiring a full cleaning of the installation next to one another, even though changing the order would have been sufficient. With every stoppage it is worth asking a simple question: why? And asking it several times in succession; this is the five-whys rule: we have a stoppage—why? Because we have to wash the installations twice? Why do you have to wash the installations twice? Because cherry has too strong an aroma. And why did you put cherry at the beginning? … and so on, up to five “why” questions.
Until the answers stop indicating the symptom and begin indicating the cause. “The line stopped”—why? “We ran out of lids”—why? “The delivery arrived in the afternoon”—why? “The order was placed too late”—why? “Because no one can see stock levels in real time”. Only this last sentence indicates what really needs to be improved.
Unnecessary transport and unnecessary movements are kilometres travelled inside the plant: a forklift carrying semi-finished product across the entire production hall because the cold buffer is on the opposite side from the line that uses it; an operator walking away forty times per shift to fetch labels.
Unnecessary stocks in food production do not merely freeze cash—they spoil.
Defects are quality complaints, batch withdrawals and a product inconsistent with the specification that must be reworked or disposed of.
Unnecessary processing means allowances: filling 1,020 g into a package declaring one kilogram “to be safe”—two per cent of raw material given away free in every item.
Delivery logistics adds its own losses. Empty kilometres—a vehicle carries goods to a distribution centre and returns empty because no one organized a return load; with a fleet of twenty articulated vehicles, this means tens of thousands of zlotys of fuel spent each year transporting air. Delays—a delivery slot at the distribution centre is missed, the vehicle waits for the next one, and goods with a short shelf life lose hours of life in the car park. And deterioration in quality during transport: every excessively long period of attaching a refrigerated trailer at the ramp, every curtain left partly open, every stop with the refrigeration unit switched off means a temperature jump that the product will not forgive.
The scale of such losses is demonstrated well by the example of a certain small delivery company. The manager noticed that couriers lost an average of seven minutes at each delivery looking for an address in an inconvenient application. Seven minutes seems like a small matter, but seven minutes multiplied by twenty deliveries a day multiplied by ten drivers is twenty-three hours a day. Almost a full-time position that simply evaporated. One correction to the application interface gave the company back an entire working day—every day. Where in your plant is there such a seven-minute loss? At the ramp? During a changeover? While filling in transport documents?
A small step instead of a grand plan
And here we reach the heart of the matter. When a loss has been identified, instinct suggests a large project: a new system, reorganization, investment. Kaizen suggests a step so small that it seems unserious—because a small step does not arouse resistance and is actually carried out, while a grand plan usually ends as a presentation. Remember one thing: a small step, but it must be performed. Small does not mean unimportant.
Instead of “we are implementing a changeover-reduction programme”—one change: changeover tools are waiting on a trolley by the line before the previous batch ends. Result: more than ten minutes less for every product change. Instead of a “raw-material waste-reduction programme”—one additional check scale and a weekly review of overfills on one, single line. Instead of a “fleet-optimization project”—one question asked while planning every route: what is this vehicle carrying on the return journey? Instead of a “new quality policy”—one additional temperature measurement at the point in the cold chain where no one has measured it before.
Each of these actions considered separately looks ridiculous against management’s ambitions. But these are the actions that take hold and, once adopted, open the way to the next ones.
PDCA: a loop, not a burst
How do we know whether a small step works? Deming also left the Japanese a tool: the PDCA cycle. Plan—plan exactly what we are improving and what result we expect. Do—implement a small change. Check—check what really changed. Act—if it works, standardize it; if not, correct it and repeat.
An example from the production hall: the objective—to reduce inter-product cleaning from sixty to forty-five minutes. The change—a new, written sequence of activities and chemicals prepared before the line is stopped. Measurement after one week: fifty-two minutes. Not forty-five, but the direction is good, and the measurement shows where the rest is leaking away: waiting for quality control to release the line. Correction: quality control receives a notification fifteen minutes before cleaning ends. Another week, another measurement. And so on in a loop—because this is a loop, not a one-off burst.
The key word is “check”. Most plants plan, many begin implementing, but almost no one stops to measure honestly what really happened. And what is not measured is not improved. The second pillar of permanence is standardization: an improvement that worked must enter the workstation instruction, the changeover checklist and the route-plan template—otherwise, it will disappear with the first holiday taken by the person who devised it.
It is also worth borrowing the practice of hansei—regularly looking back. Fifteen minutes every Friday, the entire shift team, three questions: what worked this week, what did not work, what will we do differently next week? Without looking for someone to blame—like a scientist analysing an experiment. A mistake without reflection is only a loss; a mistake with reflection is experience.
The human factor: kaizen is primarily an HR problem
Everything discussed so far—muda, PDCA, standardization—is technique. And technique is the easier part of kaizen. The harder part is people. Because systems do not report improvements, procedures do not notice that film is in the wrong place, and indicators do not come up with the idea of a tool trolley by the line. People do these things, and people can be creative or passive.
The same operator, the same planner, the same driver can be the source of dozens of ideas per year or a person who does not report a single one in ten years. The difference almost never lies in the person. It lies in what pays for them in a given organization.
A reservation is needed here: “pays” does not mean only “earns money”. People act in ways that pay for them, but the currency is not always a bonus. It may be recognition, peace and quiet, a sense of agency, and very often a desire to prove something—to oneself, the foreman or management. An employee whose idea shortened a changeover and who heard this said publicly at a briefing will look for another improvement without any bonus. An employee whose idea was stuck in a manager’s drawer or—worse—was implemented as the manager’s idea will remain silent for years. And they will be right: they were taught that creativity does not pay in this company and passivity is safe.
The greatest damage, however, is done by something else: a badly designed bonus. An incentive system always works—just not necessarily in the direction that was planned. People do not optimize what management had in mind. They optimize what you actually pay for. Two real-life examples, both authentic, demonstrate this painfully well.
Example one: a bonus for the absence of stoppages
The management of a certain cosmetics plant decided to introduce kaizen and—in keeping with the spirit of the method—began with a small step: reducing production-line downtime. The incentive for the crew was set simply: the planner and production manager received a bonus inversely proportional to the percentage of line downtime. The objective was achieved, and then some—stoppages fell significantly, the indicator on the board glowed green, and bonuses were paid.
Except that the way in which the objective was achieved ruined the company. Stoppages fell because the planner and manager did the only thing that was truly rational from their point of view: they drastically lengthened production runs. The longer the run of a single product, the fewer changeovers, cleanings and stoppages—and the higher the bonus.
The effects? The company began producing for stock, warehouses were bursting at the seams, and cash froze in inventory. Long runs killed production variety—and in cosmetics, where the market demands a broad, rapidly rotating assortment, this is a blow to the heart of the business. Formally, kaizen worked: the measured indicator improved. In reality, the company paid a bonus for making its own situation worse. One muda—waiting—was eliminated while two larger ones were cultivated in return: overproduction and unnecessary inventory.
For a food plant, this story should sound like a fire alarm because the mechanism is identical and the stakes are higher. A bonus for OEE or minimizing stoppages, set without a counterweight, will lead planning to exactly the same place: long runs of a single item. Except that in food production, overproduction does not lie calmly in a warehouse like hand cream. Overproduced yoghurt, cold meat or juice with a short shelf life means a price reduction, a return from the retail chain or disposal in a few weeks. The stoppage indicator will be exemplary, but the company will be rewarding people for producing future losses.
Example two: a bonus for fast loading
In another company, it was decided to shorten vehicle loading time. The method: rushing drivers so that they secured goods more quickly. Loading time did indeed fall. At the same time, the number of goods damaged in transport rose dramatically. The goods were OSB boards, plywood, kitchen worktops and flooring panels—in other words, goods that are destroyed in large quantities when secured badly: a shifted pallet breaks the edges of an entire stack. The company saved minutes at the ramp and paid for it with complaints, returns, repeat deliveries and the loss of customers’ trust. Bravo.
The translation to food deliveries is immediate. A rushed refrigerated-truck driver means a curtain left partly open, a pallet positioned carelessly, delicate goods beneath heavy ones, a broken cold chain because the unit “will be switched on in a moment”. Loading shortened by ten minutes can cost an entire rejected delivery—a retail chain will not accept a pallet with crushed packages or goods whose temperature is outside the specification.
Both cases share the same error, known in management as Goodhart’s law: when a measure becomes a target, it ceases to be a good measure. People were not cheating or sabotaging—on the contrary, they demonstrated creativity. The organization merely directed this creativity towards optimizing an indicator instead of improving a process. This is the key lesson: the crew’s creativity is like water; it will always flow where the incentive system has carved a channel. The question is not whether people will find ways around things—they will, and fortunately so, because that is what kaizen is about. The question is in which direction.
How should incentives be set so that they drive kaizen rather than its caricature?
Several principles follow directly from the failures above.
Firstly: never reward a single indicator in isolation—every measure must have a counterweight that blocks the simplest shortcut. Downtime, yes, but paired with inventory turnover and fulfilment of the assortment plan; loading time, yes, but paired with the transport damage and complaint rate.
Secondly: reward the entire chain, not one link—if only the planner and production manager receive a downtime bonus, the warehouse and sales will pay for their success; the indicators must bind the interest of the entire flow, from raw material to the customer’s ramp.
Thirdly: reward the reporting of problems, not their concealment—a plant in which admitting to a stoppage reduces a bonus will very quickly become a plant in which stoppages disappear from reports, not from the line.
Fourthly: remember currencies other than money—public recognition of the author of an improvement, implementation of the idea under their name, and a real influence on one’s own workstation can achieve more than a bonus, and they cannot be “optimized” against the intention.
And fifthly: before any bonus system is launched, one must ask a question that should be a mandatory point of every kaizen implementation: how can a clever person maximize this bonus without improving anything? If an answer exists—and it almost always does—the crew will find it faster than management.
Kaizen did not take off in Japan because the Japanese received bonuses for indicators. It took off because an ordinary employee on the line heard for the first time that their observation had value, that their idea would be heard, tried and—if it worked—would become a standard bearing their name. This is the essence: kaizen is not an engineering project with an HR addition. It is an HR project with an engineering addition.
What kills kaizen
Finally, a warning, because kaizen can be destroyed by several typical mistakes. Waiting for perfect conditions—“we will start after the season, after the audit, after the system implementation”—ends with never starting; done is better than perfect.
Lack of consistency—one improvement will change nothing, a thousand improvements in one direction will change everything, but only if the rhythm of weekly small steps survives the season, an inspection and a change of manager.
Failure to measure results—without numbers, every “improvement programme” is fortune-telling.
Disregarding small victories—a team which no one told that its idea shortened a changeover by ten minutes will not report a second idea.
And finally, too many changes at once—five parallel projects on one line mean that none will be completed.
Japan needed two decades to go from a global symbol of shoddy goods to a global symbol of quality. It did not do so with a single five-year plan, but with millions of small steps—and an American idea that it treated more seriously than the Americans themselves. A food plant which, instead of planning a revolution, plans one small step for this week—one question, “why are we stopped?”, one return load, one additional temperature measurement—will be a different plant in a year. Not because something great will happen. Because something small will happen every day.
Wojciech Moszczyński
Wojciech Moszczyński—a graduate of the Department of Econometrics and Statistics at Nicolaus Copernicus University in Toruń; a specialist in econometrics, finance, data science and management accounting. He specializes in the optimization of production and logistics processes. He conducts research in the area of the development and application of artificial intelligence. For years, he has been involved in popularizing machine learning and data science in business environments.
Dodaj komentarz