The History of Kanban

By Victor Hoang · Updated Oct 24, 2026

Picture a Japanese engine factory in the early 1950s. Parts are made in big batches. Each machine is expensive to set up. Every batch lands in front of the next process whether it is wanted or not. Stock piles up and storage costs rise. Nobody can see what is actually short. That's the problem the first kanban cards were built to answer.

Kanban took shape at Toyota in the 1950s. Taiichi Ohno's machine shop began letting a later process pull parts from an earlier one by sending back a card. It reached full use across Toyota's plants in 1963. It entered Western management in the late 1970s and 1980s. It became a method for knowledge work when David J. Anderson applied it to software teams at Microsoft and Corbis between 2004 and 2007.

The throughline is a signal that travels. A shop sign, a supermarket shelf, a production card and a whiteboard column each carry the same idea: let the next step in the work say what it needs, and stop making more than it asks for. This article follows that idea from its first documented form to the digital board. For the working method itself, read what kanban is; here we tell where it came from.

Timeline at a Glance

Before the Card (Edo period-1940s)

A shop sign and a slogan, long before the system

Start with the word. In Japan's Edo period, which ended in 1868, merchants hung painted kanban outside their shops. A museum review of the San Diego Mingei International Museum's exhibition says the signs advertised specific goods, and their pictorial forms enabled those who were illiterate to navigate the many storefronts. The word means, roughly, a visible board that tells you what is here.

That is a linguistic precursor, not the method. Nobody was limiting work in progress with a shop sign. What it shares with the later card is the principle that information should be placed where the person who needs it will see it.

The management precursor is also separate from the card, and it is Toyota's. Toyota's own history says Kiichiro Toyoda proposed just-in-time production when the Koromo Plant began operating, and has Eiji Toyoda recalling Kiichiro's aim: "Just make what is needed in time, but don't make too much." The same history says the early Koromo practice passed only the day's blanks to the machining shop, and that Kiichiro's system dropped transfer slips. Toyota's company site puts the Koromo body line at 1938.

Then the war intervened. In 1939, a rationing system for automobile materials made it impossible to get the right materials in the right amounts at the right time, and the just-in-time effort was eventually suspended. In the meantime the company used the Goguchi system, which grouped products into numbered units so their position could be tracked through each process. Both the Goguchi idea and the just-in-time plant layout became foundations of the later production system, Toyota says.

What this means for teams today: The first lesson is a negative one. A slogan ("make only what is needed") does not run a plant. Kiichiro had the principle and no practical signal to carry it. If your team says it values flow but has no visible way to see what is waiting, you are at the 1939 stage.

The Supermarket Idea (1940s-1954)

A shelf in a photograph becomes a way to restock a factory

The next step is the one every kanban history repeats, and the one most often told wrong. The popular version has Taiichi Ohno walking into an American grocery store and inventing kanban on the spot. The record is more careful.

Ohno's own book says a colleague who had gone to the United States came back with several photographs of a supermarket, explaining that something called a supermarket existed there. Lean practitioner Art Smalley, who traces the story through Ohno's book and Toyota's own history, dates that colleague's visit to 1951 or 1952 and Ohno's naming of the idea, "the supermarket system," to around 1952 and 1953. Ohno writes of finally visiting a supermarket in person on his 1956 trip. The Lean Enterprise Institute's glossary agrees on the photographs and says Toyota installed its first supermarket in 1953 in the machine shop of its main plant, citing Ohno's 1988 book. Eiji Toyoda, Smalley notes, called the grocery-store legend "quite flawed if you bother to check a few historical facts".

Toyota's own 75-year history gives a slightly different date and a second input. It says the supermarket method was introduced in 1954, with hints from a spring 1954 industry-publication report on a Lockheed aircraft plant in the United States, which reportedly saved $250,000 a year by using the supermarket method to assemble jet bombers. Toyota's engineers took from it a role reversal: the later process is the customer, and the earlier process is the supermarket the customer shops from. Before this, Toyota says, parts were pushed forward whether or not the next process needed them, which built up inventory and storage cost.

Two sources, 1953 and 1954. They need not conflict: one dates the first shop-floor trial, the other dates Toyota's formal record of the method. Toyota applied it first to general assembly, then to the machine assembly plant pulling from the machining plant. And it adds the step that makes this history about kanban and not only supermarkets: the supermarket method evolved into the kanban system by adding an instruction card or board showing the product name, product number and quantity.

Here the precursor ends and the method begins. A supermarket shelf tells a stock clerk what to restock. A card tells a machine operator what to make.

What this means for teams today: Notice that the breakthrough was a role change, not a gadget. The next step is the customer of the previous step. Ask of any handoff in your team: who is the customer here, and how do they tell the supplier what they need and how much?

Ohno's Shop Floor (1954-1965)

Cards spread from one machine shop to every plant, then to suppliers

This era starts when the card exists and ends when it runs a company. Toyota's history says kanban are instruction tools that show which parts are needed, where and when, and that the new management method was adopted at all plants in 1963. With it, parts moved between plants in only the quantities needed, and in-process inventory shrank. By 1965, Toyota also used kanban to retrieve parts from suppliers. The company says it explained to suppliers that adopting the same techniques would improve their own operations, and those who wished to do so introduced the system step by step.

That is about a decade from a first trial in one machine shop to a company-wide rule, and Toyota's page is plain that the system was not perfect when it spread. It describes related problems it had to solve as it went: leveling production so processes were not overburdened, installing andon alerts at the Kamigo Plant in 1966, and an online control system introduced at the Takaoka Plant in 1966 that caused assembly errors because downstream work could start early. Toyota replaced it with instruction sheets attached to vehicle bodies around 1970.

Ohno, who Toyota says established the system as a former Executive Vice President, also argued for reducing the number of workers needed instead of only saving labor. Toyota ended fixed staffing so lines could run with fewer people when demand fell. Kanban was never only a card. It lived inside a plant that levelled its volume, stopped on defects and had people willing to change staffing.

Toyota's company site describes the ideal behind all of it: "Making only what is needed, when it is needed, and in the amount needed." Kanban was the daily mechanism for that sentence. For the assembly-line background Toyota's pull system answered, see our history of the assembly line.

What this means for teams today: Copying the card without the plant around it is how kanban fails. If a team adopts a board but keeps batch handoffs, no limits and a pushed schedule, it has the sign but not the shelf. Our guide to a kanban system for manufacturing covers what the surrounding system needs.

The Lean Export (1970s-1990s)

Kanban leaves Japan inside papers, books and a famous study

Outside Toyota, almost nobody could read about the system until the late 1970s. In 1977 four Toyota managers, Y. Sugimori, K. Kusunoki, F. Cho and S. Uchikawa, published "Toyota Production System and Kanban System" in the International Journal of Production Research, described as the first detailed description of the system in English. One of the authors, Fujio Cho, later became chairman of Toyota.

Ohno's own book followed. Toyota Production System: Beyond Large-Scale Production first appeared in Japan in May 1978 and reached its twentieth printing by February 1980. The first English edition, from Productivity Press, came in 1988. Library records list it as a translation of the Japanese title Toyota seisan hōshiki, and it carries the account of the supermarket story from the man himself.

A second route into the West was practical. A 1985 Productivity Press book credited to the Japan Management Association, Kanban and Just-In-Time at Toyota, is based on seminars that Ohno and other senior Toyota production staff gave to introduce just-in-time to Toyota's own suppliers. It shows how a simple visual system can control production, which is why it travelled so well to other factories.

The third route was the label. In 1990 the MIT researchers James Womack, Daniel Roos and Daniel Jones published The Machine That Changed the World, which the Lean Enterprise Institute says thoroughly described lean thinking. After that, kanban was usually met as one tool inside "lean," next to just-in-time, jidoka and continuous improvement. Our overview of lean manufacturing principles shows where pull systems sit in that family.

Notice what happened to the meaning. In Toyota's plants, a kanban was a specific card with a specific job. In the Western export it became a concept, "pull," and sometimes a synonym for just-in-time in general. That loosening is what let the next generation borrow it for work that has no parts, no machines and no cards.

What this means for teams today: When you read a kanban claim, ask which meaning it uses: the signal card, the pull principle or the whole production system. Results from one do not automatically carry to another.

The Software Board (2004-2010)

Developers borrow the card, visualize the work and limit it

The leap from factory to software was made by people trying to fix a specific problem, not by a theory. Anderson's own history dates it to October 2004, when Dragos Dumitriu, a manager at Microsoft, asked him for help and Anderson designed a pull system for the XIT Sustaining Engineering team, a group working offshore in Hyderabad, India. In winter 2005, Anderson writes, the product-development writer Donald Reinertsen encouraged him to try kanban systems, and Anderson reframed the Microsoft implementation as a "virtual kanban system."

The practices crystallized at Corbis. Anderson joined in September 2006, and a kanban system replaced the project-centric approach to minor application upgrades. In January 2007 Darren Davis proposed a card wall on a whiteboard to visualize the workflow, and in April 2007 Corbis rolled kanban out across its project portfolio. Anderson presented the full Corbis implementation with Rick Garber at the Lean New Product Development conference in Chicago in June 2007, to about 55 attendees, and gave an open-space session at Agile 2007 in Washington, DC, to about 25.

So the second turning point was the fusion of two older things. From Toyota came pull and the limit on work in progress. From the agile world, which had been named at Snowbird in 2001, came the habit of visualizing work in a team space. The result was less a process than an overlay. It is applied to the workflow a team already has instead of replacing it.

Spread followed the usual route of a practitioner movement: talks, a discussion group and a conference. Anderson's history records that Karl Scotland, Joe Arnold and Aaron Sanders asked him to set up the kanbandev discussion group after the 2007 Agile conference, that he started a consulting company in September 2008 with Dan Vacanti as first associate, and that the first Lean Kanban conference was held in Miami in May 2009. Anderson's book followed in May 2010, published by Blue Hole Press, and the Kanban University guide to the method still names it as its basis.

Compare this with the Toyota arc. The Toyota system moved inside one company over a decade. The software version moved through conferences and books in about six years, because there was no plant to rebuild, only a workflow to make visible. In an interview on how kanban took off, Anderson is reported as finding two groups of adopters: companies that had struggled with agile methods, and companies that had never tried to become agile. The second group proved more receptive because kanban does not require changing anyone's job. For how the sprint-based alternative differs, see Scrum versus Kanban, and for the older sequential model that agile reacted against, our history of the waterfall model.

What this means for teams today: The software version asks very little at the start: draw your workflow, make the work visible, limit what is in progress. You can begin tomorrow. The hard part is the second and third steps, deciding the limits and acting when they bite. Our guide to WIP limits covers how.

The Digital Board (2009-present)

The board moves to individuals, tools and the wider profession

This era overlaps the last one. While Anderson was building the method for teams, Jim Benson was shrinking it to one person. According to Anderson's account, Benson launched Personal Kanban as a series of blog posts in July 2009, after he, Corey Ladas and Anderson had formed Modus Cooperandi in January 2008. The Personal Kanban site, run by Benson and Tonianne DeMaria, sums up its approach as "Visualize Your Work | Limit Your Work-in-Progress". Two rules, no software required.

Since then the board has gone digital, and the idea has been standardized. The official guide from Kanban University says plainly that "Kanban is not a methodology nor a process framework": it is a management method applied to an existing way of working. It defines a WIP limit as the maximum number of work items allowed at one time, and names lead time, delivery rate and WIP as the core metrics. That is a long way from a Toyota card, yet two things are unchanged. The signal is visible, and the amount of work in the system is capped.

The word has also drifted again. Most people today say "kanban board" for any set of columns with cards, whether or not anything is limited, and the same cards appear inside tools built for every kind of work. The history suggests the test: if nothing is capped and nothing is pulled, it is a task board, not kanban. Hybrid setups that borrow the board inside a sprint, such as Scrumban, and mixed-method teams described in our history of hybrid project management show the same choice at the team level.

What this means for teams today: Whatever tool you use, check the two things the history says make it kanban. Is the work visible? Is the amount in progress limited, with an agreed response when the limit is hit?

Key People

Key Facts

Key Facts: Kanban History

• Toyota's history dates the supermarket method to 1954 and says it became the kanban system by adding an instruction card.
• Kanban was adopted at all Toyota plants in 1963 and used to retrieve supplier parts by 1965.
• The first detailed English description of the system appeared in 1977; Ohno's book first appeared in Japan in May 1978.
• Anderson's history dates the first software pull system to October 2004 and the Corbis rollout to April 2007.
• The official guide says Kanban is not a methodology nor a process framework.

The Pattern Behind the History

We call this the Signal and Limit pattern. It is our reading of the timeline, not a research result.

1. A problem comes before the tool. Excess stock at Toyota, stalled software changes at Microsoft: the card answered something that hurt.
2. A borrowed picture carries the idea. A supermarket shelf in a photograph, a card wall on a whiteboard. The image travels faster than the theory.
3. The next step becomes the customer. Each version turns the receiving step into the one that asks.
4. A limit makes the signal real. Stock capped by card count, work capped by WIP limit. Without a cap, it is only a display.
5. The name outruns the method. A shop sign, a production card, a management system, a software method and a column of sticky notes have all been called kanban.

What It Teaches Teams Today

• Fix the receiving step first. Let the next step in your work say what it needs and how much, and treat it as the customer.
• Cap work in progress and agree in advance what happens when the cap is hit.
• Expect the benefits to build slowly. Toyota took about a decade to move kanban from one shop to every plant and then to suppliers.
• Do not copy the board without the surrounding discipline: levelled volume, quick response to problems, and willingness to change how people are assigned.
• Be precise in language. Say "task board" if nothing is limited, and "kanban" when work is pulled and capped.

For the wider story, read the history of project management, the evolution of project management methodologies and our history of the Gantt chart, another visual tool in this collection's story.

Kanban's long record shows one stubborn idea: work flows better when the people downstream can say what they need, and the people upstream are not allowed to bury them. The card has changed material several times. The idea has not.