Kanban (かんばん) and the pull system are the signaling and control core of Toyota's Just-in-Time production. The records below state what a kanban actually is, how the production and withdrawal cards link processes through a supermarket loop, where the method came from inside Toyota, and how Toyota uses card count as a deliberate improvement tool — drawn from Ohno, the 1977 Sugimori paper, Toyota's own 1970 casebook and Engineering Society materials, and Toyota OB accounts.
Kanban is correctly written in hiragana, not katakana or kanji
A Toyota OB who joined the kanban work in 1964 states the proper spelling is かんばん in hiragana. It is not a foreign loanword, so katakana is wrong, and it is more than a mere signboard, so the kanji 看板 misses the meaning. Toyota writes it in hiragana because the word names a whole production system the company created, not just a physical card.
Toyota OB memoir, トヨタ生産方式(かんばん方式)の裏話 — artoflean.com/reference/kanban
Kanban means a reusable instruction sheet, not a single-purpose card
In the 1970 internal definition, a kanban is a repeatedly usable instruction sheet that consolidates the content of a production order, an issue order, a conveyance or withdrawal order, a purchase order, and a delivery instruction onto one card. By carrying all that information, it removes the need for separate ledgers and slips, simplifies clerical work, and reduces staffing. The card is an information-carrying medium, not decoration.
かんばん事例集 (Toyota head-office Kanban casebook, March 1970) — artoflean.com/reference/kanban
Kanban is one JIT tool, not the whole system
Kanban is a signaling and control mechanism that authorizes production or withdrawal within a pull system. It is one important JIT tool, not the whole of JIT. JIT also requires stability, flow, takt, leveling, quality, changeover, and management response.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/kanban
A pull system authorizes work from actual downstream need, not local convenience
A pull system authorizes production or movement based on actual need from the customer or next process. Pull is not an abstract preference over push; it is the synchronization logic that stops each process from making what is convenient for itself rather than what the next process needs. The practical test is whether the process is supplying the right item, in the right quantity, at the right time, in the right sequence, to the right location.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/pull-system
The full Toyota term for pull is subsequent-process withdrawal
The pull principle is 後工程引き取り (atokoutei hikitori), "subsequent process withdrawal." The next process in line reaches back and takes what it needs. The 1970 casebook states the principle directly: the subsequent process goes to the preceding process and takes the necessary things, in the necessary quantity, at the necessary time, and the preceding process responds with the minimum necessary inventory.
Toyota 75-Year History; かんばん事例集 (1970) — artoflean.com/reference/pull-system
Hikitori combines the verbs to pull and to take
The word hikitori (引き取り) combines 引き (hiki, to pull) and 取り (tori, to take). It is an ordinary Japanese word meaning to withdraw, retrieve, or take back. In TPS it describes the downstream process pulling and taking parts from upstream — the customer reaching back to the shelf.
Pull System encyclopedia entry — artoflean.com/reference/pull-system
A supermarket is designed, controlled inventory, not a warehouse with a new name
A supermarket is a controlled inventory location used to connect processes through replenishment pull when continuous flow is not practical. It must have defined items, locations, quantities, replenishment signals, ownership, and abnormal-response rules. It exists because two processes cannot yet be directly flowed or sequentially controlled under current conditions. A supermarket without a clear pull signal, standard quantity, owner, or abnormal response is just a warehouse.
Art Smalley, Art of Lean (JIT skill pack); Supermarket encyclopedia entry — artoflean.com/reference/supermarket
Mizusumashi means water spider and describes the looping route
Mizusumashi (水すまし) refers to the whirligig beetle, a small aquatic insect that moves rapidly in circular patterns across the water surface. The name was adopted because the material handler, like the beetle, circles the same route repeatedly. The route is documented as standardized work, and the timed cycle is typically twenty to sixty minutes depending on layout and consumption rate.
Mizusumashi encyclopedia entry — artoflean.com/reference/mizusumashi
Flow means continuous movement with minimal waiting and batching
In JIT, flow means designing work so material, information, or service moves through the process with minimal waiting, batching, backtracking, and hidden inventory. The Japanese word nagare (流れ) is the everyday term for a flowing stream; nagare seisan (流れ生産) means flow production. Flow shortens lead time and makes abnormalities visible, because work that waits in batches, queues, or warehouses hides problems.
Art Smalley, Art of Lean (JIT skill pack); Flow encyclopedia entry — artoflean.com/reference/flow
Flow work means value is added as the item flows, not mere conveying
Ohno distinguished flow work (流れ作業) from merely flowing work (流し作業). In flow work, the item is processed and gains value at each process while it flows; simply conveying it on a conveyor is not flow work but flowing work. Ohno's appendix glossary makes this distinction explicit as a guard against mistaking movement for value-adding flow.
Ohno, Toyota Production System (1978) — artoflean.com/reference/flow
The kanban system originated as the supermarket method
Toyota's pull mechanism was first called the supermarket method (スーパーマーケット方式). It was adopted at the head-office machinery plant and later, with the addition of cards showing part number and other information, became known as the kanban method. The supermarket logic — the preceding process is the store, the following process is the customer — is the conceptual core that the cards operate.
Toyota Engineering Society, かんばん方式解説書 (自動車の知識) — artoflean.com/reference/supermarket
Ohno credited the American supermarket he saw on a 1956 visit
Ohno wrote that on a 1956 visit to America his strongest impression was the spread of supermarkets — stores where the customer gets the needed goods, when needed, in the needed amount. He took the hint of seeing the supermarket as the preceding process in a production line: the customer (following process) buys the needed part, when needed, in the needed amount, and the store (preceding process) immediately replenishes only what was withdrawn.
Ohno, Toyota Production System (1978) — artoflean.com/reference/supermarket
Ohno reframed the conventional flow by reversing its direction
Ohno described his method as "anti-common-sense": instead of the preceding process supplying parts to the following process, the following process goes to the preceding process to withdraw what it needs, when it needs it, in the needed amount, and the preceding process makes only the amount withdrawn. His rationale was that only the final assembly line accurately knows the necessary timing and quantity, so it pulls upstream and each process produces only what was taken.
Ohno, Toyota Production System (1978); Sugimori et al. (1977) — artoflean.com/reference/pull-system
The supermarket method dates to 1954 and became the kanban method in 1963
The 1970 casebook records that the supermarket method was introduced as a production-control method at the head-office machinery plant in 1954, developed into the kanban method in 1963, and kept evolving afterward. The casebook states the method was adopted from 1963 onward and progressively implemented case by case in the head-office plant until it reached full adoption.
かんばん事例集 (1970) — artoflean.com/reference/kanban
Internal kanban diffused over more than ten years before reaching suppliers
Ohno wrote that even Toyota's internal kanban took more than ten years to establish, and called it a precious experience. A flow between final assembly and machining formed around 1950 with modest synchronization, kanban became possible company-wide around 1962, and only after that were cooperating firms invited to observe and learn. The 1977 Sugimori paper states the system had been rooted in Toyota for about twenty years by then.
Ohno, Toyota Production System (1978); Sugimori et al. (1977) — artoflean.com/reference/kanban
Sugiura describes the kanban system as fully introduced in 1965
Mikio Sugiura describes the kanban system being tested in machinery plants, then expanded, and fully introduced in 1965. In that expansion, cards were attached to transport boxes, the number of kanban was adjusted according to supplier distance and delivery pattern, line-side inventories dropped sharply after trial and error, and synchronization of upstream and downstream operations improved.
Mikio Sugiura, via Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/kanban
Kanban operationalizes Kiichiro Toyoda's earlier just-in-time intent
Kanban should be treated as Ohno's postwar development mechanism, not as the origin of JIT itself. Eiji Toyoda describes Kiichiro Toyoda's 1938 flow-production thinking as existing before kanban: at the Koromo plant Kiichiro wanted each day to make only the necessary things in the necessary quantities, using daily slips, with no kanban yet. Eiji Toyoda states that Ohno revived and developed after the war, using kanban, what had once died.
Toyota 75-Year History; Art Smalley, Art of Lean — artoflean.com/reference/just-in-time
Ohno took up "just in time" in the machine plant in 1954 amid tolerated chaos
Sugiura adds that when Ohno became an executive manager in 1954 he took up the challenge of manufacturing just-in-time in the machinery plant. The constant changes threw the plant into chaos, but Taizo Ishida and Eiji Toyoda let him continue. That senior tolerance for disruption during deliberate capability building is part of the history, not a side note. Ohno himself credited the big-hearted top management who left the work entirely to him.
Mikio Sugiura, via Art Smalley; Ohno, Toyota Production System (1978) — artoflean.com/reference/taiichi-ohno
Ohno reorganized machining from process layout to process-sequence flow
Around the time of the Korean War in 1950, as machine-shop manager of the Koromo plant, Ohno changed the equipment layout from processing in big batches and sending to the next process, to arranging different machines in process order and making one piece at a time. This required multi-skilled workers, so there was much resistance, and many problems arose because machines were not set to stop at completion and adjustment elements required skill.
Ohno, Toyota Production System (1978) — artoflean.com/reference/one-piece-flow
The 1977 paper is the first authoritative English account of TPS and kanban
The Sugimori paper, authored by Toyota's own Production Control Department including F. Cho, later Toyota chairman, is the first authoritative description of TPS and the kanban system in the English-language academic literature. It was presented at the 4th International Conference on Production Research in Tokyo in August 1977 and predates Ohno's English book of 1988 and Monden's textbook of 1983. For historical attribution it is the canonical 1977 primary source on kanban.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/kanban
Do not credit Shingo or Ford with inventing Toyota's pull system
Toyota's JIT and kanban were internally developed production control, not the invention of Shigeo Shingo, who taught industrial engineering and setup-reduction methods and influenced many practitioners. Treat SMED and quick changeover as enablers of pull, not its origin. Likewise, Ford had assembly flow and earlier industries had parts standardization before Toyota; Toyota's pull system is its specific production-control and improvement system, not a claim to have invented all flow.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/shigeo-shingo
A kanban must answer six practical control questions
A working kanban answers what item, what quantity, from where, to where, by what timing or rule, and under what abnormal-response condition. A card, bin, empty space, or electronic signal can serve the function, but the mechanism must match the production-control logic behind it. Designing a card before understanding the replenishment logic is a common failure.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/kanban
The two primary kanban types are production and withdrawal
The classic distinction is between the withdrawal kanban, which authorizes movement of material from a supplying location to a consuming location, and the production kanban, which authorizes production to replace what was withdrawn. The 1977 Sugimori paper names these the conveyance kanban and the production kanban, both always attached to the containers holding parts.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/kanban
The Japanese term for the production kanban is shikake kanban
The production instruction kanban is 仕掛けかんばん (shikake kanban). 仕掛け conveys "to set in motion" or "mechanism," so the term means the kanban that sets production in motion. It travels between a process and its output supermarket and returns to the producing process as authorization to make replacement parts.
Production Instruction Kanban encyclopedia entry — artoflean.com/reference/production-instruction-kanban
The Japanese term for the withdrawal kanban is hikitori kanban
The withdrawal kanban is 引き取りかんばん (hikitori kanban), from 引き取り, "to take back" or "to withdraw." It authorizes retrieval of a specific part in a specific quantity from a supermarket or upstream location to a downstream process. In some Toyota documentation it is also called the conveyance kanban (運搬かんばん, unpan kanban).
Withdrawal Kanban encyclopedia entry — artoflean.com/reference/withdrawal-kanban
The two-card system separates authorization to make from authorization to move
In the earliest experiments a single card could serve both functions, but as the system extended across more processes and to suppliers, the two-card system became standard. The production cycle and the delivery cycle run on different rhythms and between different locations, so separate cards keep production authorization independent of material movement. Without that separation, the temptation to overproduce increases.
Production Instruction Kanban encyclopedia entry — artoflean.com/reference/production-instruction-kanban
The card exchange at the supermarket is where the two kanban loops connect
When a downstream process empties a container, the freed withdrawal kanban and the empty container are carried to the supermarket. There the withdrawal kanban is swapped onto a full container, and the production kanban detached from that container is left at the collection post to signal upstream replenishment. This physical card swap is the mechanism that links downstream consumption to upstream production authorization.
Withdrawal Kanban encyclopedia entry — artoflean.com/reference/withdrawal-kanban
The signal kanban manages batch processes within a pull system
A signal kanban is a triangular production kanban (三角かんばん, sankaku kanban) used where changeover time makes one-piece flow impractical, such as stamping or molding. It is placed at a calculated reorder point inside the supermarket stack; when consumption draws inventory down to that point, the exposed triangle authorizes the batch process to change over and produce a replenishment lot. This maintains pull even in a batch environment.
Signal Kanban encyclopedia entry — artoflean.com/reference/signal-kanban
The signal kanban is also called the lot-making kanban
Some Toyota literature calls the signal kanban the lot-making kanban (ロット作りかんばん) because it authorizes production of a full lot rather than a single container. The 1970 casebook describes the same idea for lot-processed parts: when stock falls to the lead-time quantity, produce the next fixed lot, and a lower-limit inventory must be set.
Signal Kanban encyclopedia entry; かんばん事例集 (1970) — artoflean.com/reference/signal-kanban
Toyota practice distinguishes three pull types
Toyota practice distinguishes replenishment (supermarket) pull, sequential pull, and mixed pull, and this terminology is preferred over generic "pull" talk. Replenishment pull, where a downstream process withdraws from a controlled supermarket and the withdrawal triggers upstream replenishment, is often the most common starting point. Sequential pull triggers production in a specific order without a large finished-goods supermarket and requires stronger stability and schedule discipline.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/pull-system
Mixed pull combines replenishment and sequential logic
Mixed pull uses both supermarket and sequential mechanisms when product variety, lead time, demand pattern, or process constraints require more than one approach. The encyclopedia describes the common split as supermarket logic for high-volume parts, perhaps 80 percent of volume, and sequential logic for low-volume items, perhaps 20 percent.
Pull System encyclopedia entry; Art Smalley, Art of Lean — artoflean.com/reference/pull-system
Supplier kanban is a form of withdrawal kanban
Supplier kanban, used between Toyota and outside suppliers, is a form of withdrawal kanban. The card travels to the supplier as authorization to deliver a specific quantity to a specific receiving dock on a specific schedule, and defines the logistics of where to get parts, how many, and where to deliver.
Withdrawal Kanban encyclopedia entry — artoflean.com/reference/withdrawal-kanban
The mizusumashi is the physical link in the kanban loop
A mizusumashi is a dedicated material handler who follows a fixed, standardized route on a timed cycle to supply parts and collect empty containers and kanban cards. When an operator consumes parts, the freed kanban goes to a collection box; the mizusumashi picks up these cards, which authorize withdrawal of replacement parts from the supermarket. The role connects supermarkets to the point of use and keeps operators from leaving their stations.
Mizusumashi encyclopedia entry — artoflean.com/reference/mizusumashi
Small, frequent delivery by cart replaces large forklift batches
The mizusumashi typically uses a small towing cart pulling a train of carts through narrow aisles, not a forklift. This delivers small quantities frequently rather than large quantities infrequently, directly supporting the just-in-time principle. Using forklifts encourages batch delivery, requires wide aisles, and creates safety hazards.
Mizusumashi encyclopedia entry — artoflean.com/reference/mizusumashi
The 1970 casebook required frequent small-lot conveyance and mixed loads
The Toyota Engineering Society explanation states that because conveyance moves only the amount whose kanban have come off, transport must be small and frequent. Consolidating kanban from many places into mixed loads removes conveyance loss while still achieving small-lot withdrawal, so processes stay connected with little stock on hand. It also notes the supermarket store should be wide across the front (one slot per part type) and shallow in depth.
Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/transport
No kanban, no production is the fundamental supermarket rule
The fundamental rule of the supermarket system is that nothing is produced without a withdrawal signal. Every container in the supermarket must carry a kanban, and every withdrawal must generate a replenishment signal. Parts without kanban, or kanban without corresponding parts, indicate a system breakdown, and producing just-in-case or to keep busy breaks the pull system and lets inventory grow uncontrolled.
Supermarket encyclopedia entry — artoflean.com/reference/supermarket
Supermarket size is calculated, not arbitrary, and should shrink over time
The maximum inventory in a supermarket is calculated from upstream replenishment cycle time, downstream demand pattern, lot sizes where changeovers are required, and a safety factor for variability. Toyota continuously works to reduce supermarket inventory by improving upstream cycle times, reducing changeover, and increasing delivery frequency. The supermarket quantity is a measure of the system's capability, and a supermarket that stays the same size for years indicates stagnation.
Supermarket encyclopedia entry — artoflean.com/reference/supermarket
The pacemaker is the single point where the schedule enters the value stream
A pacemaker is the point in the value stream where the schedule is normally introduced and from which upstream production is controlled by pull. Downstream of the pacemaker, flow should be as continuous as practical; upstream, pull loops and supermarkets regulate replenishment. The pacemaker is usually the most useful point to schedule because its output most directly sets the rhythm of the value stream.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/material-and-information-flow-analysis
Too many schedule points turn a system into push with lean labels
Together, supermarkets and pacemakers prevent every process from being scheduled independently: schedule one point, pull upstream where possible, flow downstream where practical. Choosing the wrong pacemaker creates confusion, and if many points are independently scheduled the system often becomes push with lean labels. Each department then follows its own schedule and sub-optimizes the whole.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/material-and-information-flow-analysis
Flow demands multi-process handling, distinct from multi-machine handling
Ohno distinguished multi-machine handling, where one worker tends several lathes, from multi-process handling, where one worker handles several different processes such as a lathe, a mill, and a drill in sequence. One-piece flow in a cell requires multi-process handling, which broke the entrenched view of workers as single-skill specialists. Ohno noted this was impossible in America under craft unions but doable in Japan with the will.
Ohno, Toyota Production System (1978) — artoflean.com/reference/standardized-work
One-piece flow is the ideal; small-lot flow is the common realistic state
One-piece flow is the ideal form of flow where practical, and small-lot flow is often the realistic next condition when changeover, distance, equipment constraints, or product mix make one-piece flow impractical. Small lots are not failure; they may be the correct development path while changeover time, quality risk, material sensitivity, or equipment constraints are improved progressively. The TPS glossary defines one-piece flow (一個流し, ikko nagashi) as processing and passing one piece at a time to the next process.
Art Smalley, Art of Lean; Toyota TPS glossary via gemba-coach draft — artoflean.com/reference/one-piece-flow
Sugimori states the goal as one-piece production and conveyance everywhere
The 1977 Sugimori paper describes the approach as every process producing one piece, conveying one at a time, and holding one piece in stock between equipment and processes, with no process overproducing or holding surplus. This is enabled by greatly shortened setup times for smaller lots, multi-purpose machine layout along the process sequence, and repetitive mixed-load conveyance extended to subcontractors.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/one-piece-flow
Shorter changeover is what makes small lots and pull practical
Shorter changeover makes smaller lots practical, and smaller lots make JIT easier by reducing waiting, improving responsiveness, supporting leveling, and exposing problems sooner. Changeover work is an enabler of JIT, not a standalone trophy; the point is improved synchronization with customer and next-process need, not a dramatic one-off setup-reduction event. Reducing lot size by command, without improving changeover, quality, material supply, scheduling, and response, produces downtime, shortages, and expediting.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/smed
Leveling the final assembly line is the precondition for kanban to work
The 1977 Sugimori paper states that if subsequent-process withdrawals vary widely, upstream processes and subcontractors must hold peak capacity or excess inventory, so the prerequisite is to level the final assembly line. Final lines are mixed-product lines; daily production is the monthly schedule by specification divided by working days, and within a day each specification is sequenced to appear at its own cycle time. Leveling the final line levels all upstream processes too.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/heijunka
Internal leveling is a precondition for applying supplier kanban
The 1970s Toyota Engineering Society explanation states plainly that if a process makes a batch for its own convenience and then withdraws that amount upstream, the preceding process must hold surplus equipment and people to avoid shortages. Therefore kanban requires leveled production and leveled withdrawal, and supplier kanban cannot be applied until the company's own internal production is leveled.
Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/heijunka
Kanban quantity expresses policy, not a fixed given
Card quantity expresses policy: demand, replenishment time, container size, supplier or process distance, delivery frequency, variation, and risk assumptions. Sugiura's supplier example matters because daily delivery and every-other-day delivery do not require the same kanban logic or truck-loading behavior. Treating card count as decoration misses that the number encodes the system's assumptions.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/kanban
The 1977 kanban formula treats lead time and inventory as policy variables
Sugimori's 1977 paper gives the number of kanban as demand per unit time times the sum of waiting time and processing time, times one plus a policy factor alpha, divided by container capacity. Container capacity is held to no more than ten percent of daily requirement and alpha to no more than ten percent. The crucial point is that these factors are not accepted as given; they are targets for positive improvement.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/kanban
Holding kanban count fixed as demand rises forces lead-time reduction
Sugimori's notes explain that the number of kanban is deliberately held rather fixed despite variation in demand, so when demand rises a shop must cut processing and waiting time, that is, lead time. A weak shop copes by temporarily raising alpha, adding kanban, and top managers read alpha as an indicator of a shop's improvement capability. When demand falls, lead time grows relatively and idleness becomes visible, prompting headcount reduction.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/kanban
The 1970 casebook lists four improvement rules for kanban
The Toyota Engineering Society explanation states four rules that turn kanban from a production tool into an improvement tool: do not convey or make without a kanban; always attach a kanban to the physical item; require 100 percent good product; and reduce the number of kanban over time. Following these rules prevents overproduction, surfaces defects so recurrence prevention is forced, and shrinks buffers until instability must be solved.
Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/kanban
Ohno's six kanban roles and rules of use
Ohno's appendix tabulates kanban's roles and matching rules: it carries withdrawal information, so the following process withdraws only as many as kanban have come off; it carries production information, so the preceding process makes the detached items in the order detached and only as many as detached; with no kanban, do not convey or make; a kanban must always be attached to the physical item; the product must be 100 percent good so a defect-making process feels the pain; and reducing the number of kanban surfaces problems and manages inventory.
Ohno, Toyota Production System (1978) — artoflean.com/reference/kanban
Kanban abolished the schedule chart for upstream processes
The Toyota Engineering Society explanation states that once the following process withdraws and the preceding process makes only what was taken, the upstream schedule chart becomes unnecessary, even an obstruction, and was abolished. Conventionally a schedule was issued to every process; under kanban the plan is given only to the final process, and upstream processes simply replenish what their kanban detach. Free of plan constraint, each process automatically absorbs production variation within a band, so kanban acts as a fine-adjustment device.
Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/pull-system
The sequence plan goes to one place only — the final assembly line
Ohno describes the Toyota information system sending the leveled sequence plan to one place only, the head of the final assembly line, and calls this a great feature of the system. Each sub-assembly process combines with the main line and is told only the information it needs when it needs it; more information than that is, for now, unneeded. Sending detailed plans to every process would require vast peripheral equipment, be costly and unreliable, and induce running-ahead and overproduction.
Ohno, Toyota Production System (1978) — artoflean.com/reference/pull-system
The 1970 casebook set explicit rules for designated part and kanban locations
The 1970 casebook specifies that each part must have a defined storage address and never be placed anywhere else, with locations chosen for conveyance convenience and quality preservation, and the part's kanban marked with both upstream and downstream addresses. It mandates kanban posts at fixed positions for the cards a process detaches, and a separate red post at the preceding process to signal an expedite condition when a withdrawing process finds no parts available.
かんばん事例集 (1970) — artoflean.com/reference/visual-management
The 1970 casebook defined when an operator may stop the line
The 1970 casebook lists the conditions under which a worker stops the line: when a defect occurs in the process, when standardized work cannot be kept, when only the kanban quantity is complete, when an upstream quality defect is found, and when a shortage is about to occur. It frames line stops as making process problems visible, to be used 100 percent for improvement, and assigns supervisors the duty of countermeasures so the line need not be stopped again for the same cause.
かんばん事例集 (1970) — artoflean.com/reference/fixed-position-stop-system
Lost kanban must trigger responsibility tracing, not casual reissue
The 1970 casebook treats the kanban as the most important element of the system and warns that a lost kanban can halt production and make the target unachievable. When a lost kanban is discovered, the responsible area must be identified and made to take recurrence-prevention measures before the lost card is reissued. A Toyota OB notes that casual replacement of supposedly lost cards is one way card counts inflate and inventory grows.
かんばん事例集 (1970); Toyota OB memoir — artoflean.com/reference/kanban
A kanban template is meaningless without its control information
If asked to make a kanban template, the right first step is to explain the information the kanban must control rather than hand over a generic card. A usable card carries item, quantity, source, destination, trigger, timing, and response rules. The 1970 casebook and a Toyota OB list the actual fields used at Toyota: part number and name, box type and capacity, preceding and following process, plus a large sorting number called the back number.
Art Smalley, Art of Lean; Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Kanban is best introduced between two controlled processes first
The encyclopedia advises beginning kanban between two processes you control, not across the whole value stream, using physical cards, setting initial quantities conservatively, and reducing them over time. Problems will surface and should be solved rather than buried by adding cards back. Extension to suppliers should come only after internal pull is stable, matching Toyota's own multi-year internal-first sequence.
Kanban encyclopedia entry — artoflean.com/reference/kanban
A fixed-capacity cart can itself serve as the kanban
Ohno notes that for withdrawing unit assemblies such as engines and transmissions at final assembly, a fixed cart that holds only a set amount is used, and the cart itself doubles as the kanban. The principle is to attach a kanban, but the preceding and following processes can agree withdrawal rules — how many carts and so on — and get the same effect with a simple number tag. This shows the kanban function can be carried by any disciplined signal, not only a paper card.
Ohno, Toyota Production System (1978) — artoflean.com/reference/kanban
Wrong container quantity is a common lot-size failure in kanban design
A recognized failure mode is choosing container quantity for convenience rather than for flow, changeover, demand, and handling logic. The container quantity is the lot size embedded in the pull loop, and Sugimori's formula holds container capacity to no more than ten percent of daily requirement. Designing the card before deciding the right container quantity puts the cart before the horse.
Art Smalley, Art of Lean (JIT skill pack); Sugimori et al. (1977) — artoflean.com/reference/kanban
Push is a synchronization failure, not the enemy in an ideology contest
Push occurs when a schedule or process self-optimizes and makes more than the customer or next process requires, or delivers too much, too little, too early, or too late. It is commonly created by local efficiency targets, OEE pressure, narrow theory-of-constraints thinking, batch economics, or schedule adherence that ignores downstream need. The local process can look efficient while the total system is worse. Treating push versus pull as an ideology contest misses that push is usually a synchronization problem.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/pull-system
Kanban equals pull is a misconception
Kanban may operate a pull system, but cards alone do not make a system pull. The common error is to install cards or electronic signals and call the result a pull system while ignoring stability, lot size, quality, leveling, and management response. Saying kanban equals JIT is the same mistake at a larger scale.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/pull-system
Pull without leveling makes signals oscillate wildly
A common failure is pull without leveling, where signals fluctuate because the pacemaker or schedule is not leveled enough. The 1970 casebook warns that uneven demand at the final process appears as load swings upstream, and the further upstream a process is, the larger the swing it absorbs, causing confusion and inventory excess or shortage. The casebook states that because the method aims at fine inventory adjustment, causing very large fluctuations is the thing most to be guarded against.
かんばん事例集 (1970); Art Smalley, Art of Lean — artoflean.com/reference/heijunka
JIT does not mean zero inventory; supermarkets are explicit designed inventory
JIT does not mean no inventory. Supermarkets are explicit, designed inventory, and the issue is whether inventory is positioned, sized, controlled, and improved according to system logic. Useful inventory includes cycle stock, buffer stock, and safety stock; wasteful inventory hides poor synchronization, unstable processes, excessive lot sizes, unreliable suppliers, or weak response discipline.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/inventory
A supermarket can be a JIT mechanism or a violation depending on control
A supermarket does not violate JIT when it is controlled, visible, and tied to replenishment pull. It violates JIT when it becomes unmanaged storage or a place to hide overproduction. The distinction is whether the inventory is positioned, sized, owned, and improved by system logic, or merely accumulated and renamed.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/supermarket
Hidden push uses the supermarket as a dumping ground
A recognized failure mode is hidden push, where an upstream process produces ahead despite kanban rules and uses the supermarket as a dumping ground. This breaks the pull discipline while keeping the lean labels. The casebook's rule that a process produces only the number of detached kanban and no more is the direct guard against it.
Art Smalley, Art of Lean (JIT skill pack); かんばん事例集 (1970) — artoflean.com/reference/supermarket
Kanban is not the same as software Kanban boards
The manufacturing kanban is a production-authorization signal tied to physical parts and the pull principle. David Anderson's Kanban method for software development borrowed the name but is a fundamentally different thing, a workflow-visualization method, not a production-authorization signal. Conflating the two is a common error.
Kanban encyclopedia entry — artoflean.com/reference/kanban
Electronic kanban is not automatically better than physical cards
An electronic signal can be fast and useful, but it can also hide the physical condition, bad lot sizing, poor ownership, or unstable replenishment logic. Software masking is a recognized failure mode where electronic kanban obscures rather than clarifies the physical flow and responsibility. The encyclopedia advises starting with physical cards because the discipline of handling them teaches the system's logic.
Art Smalley, Art of Lean (JIT skill pack); Kanban encyclopedia entry — artoflean.com/reference/kanban
Toyota's pull system is heavily IT-integrated, not purely manual
The pull-system encyclopedia, drawing on Art Smalley's documentation, describes a sophisticated kanban flow far beyond the paper card: assembly workers pull cards on first use, cards are collected on a timed basis signaled by the andon board, fed into an automatic sorting machine that reads bar codes and relieves inventory, the scan triggers supplier payment, and an electronic transfer prints the supplier kanban into a heijunka box. Art has criticized the view that the only answer is to unplug computers and do everything manually; Toyota integrated IT into its pull system for decades.
Art Smalley, "Sample Toyota Kanban Flow to Supplier" (2010); Pull System encyclopedia entry — artoflean.com/reference/pull-system
Kanban given to suppliers without changing one's own making becomes a weapon
Ohno warns against powerful makers telling suppliers to bring parts JIT without changing their own way of making, using kanban only to withdraw outside parts. In that case kanban turns instantly into a weapon, an alien thing far from its true role. Art's skill material echoes this as supplier dumping, where a customer claims to be lean while pushing variation and inventory burden onto suppliers.
Ohno, Toyota Production System (1978); Art Smalley, Art of Lean — artoflean.com/reference/kanban
JIT translates to other industries only by translating the logic, not the artifacts
JIT was built for Toyota production and can be translated to healthcare, service, logistics, food, and information work, but only by translating the synchronization logic, not by copying kanban cards, heijunka boxes, or takt math. The translation question is what must arrive, where, in what quantity, at what time, by what signal, and with what response if abnormal. Some systems are fixed-time with variable quantity, others fixed-quantity with variable time, and many are hybrids.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/just-in-time
Kanban is a way of managing that presupposes a flow-based way of making
Ohno states that TPS is a way of making and the kanban method is a way of managing, and that without first making production flow you cannot suddenly do kanban. When Toyota first tried kanban on the final assembly line and went to withdraw from the upstream machine shop, it was hopeless; a kanban-bearing flow had to be built going upstream process by process. Leveling as far as possible and following standardized work were the other preconditions.
Ohno, Toyota Production System (1978) — artoflean.com/reference/flow
A 1954 trade-paper article on a Lockheed plant catalyzed the supermarket method
In spring 1954 a Japanese trade paper carried a small article reporting that Lockheed in the United States had adopted a supermarket method for jet-aircraft assembly and saved 250,000 dollars in one year. The 1970 casebook records that the person who noticed this unremarkable article was Taiichi Ohno, who had been practicing Kiichiro Toyoda's just-in-time idea in the machinery plant and had already achieved partial planned flow production.
かんばん事例集 (1970), quoting Toyota's 50-year history — artoflean.com/reference/supermarket
The earliest documented trigger was a request to hold parts upstream
A Toyota OB recounts a pre-kanban back-story from the head-office plant, where two executives split responsibility for the preceding and following processes. The downstream executive, unable to keep to plan and short of storage space, asked the upstream executive to leave finished parts in place and let him come collect them when he caught up. The OB reports the upstream executive recalled this request as a likely seed for Ohno's later subsequent-process-withdrawal idea.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/pull-system
The first written documentation of the kanban method was the 1970 casebook
The first time the kanban method was put into writing was the かんばん事例集 (Kanban Casebook) compiled by the head-office engineering department in March 1970, more than fifteen years after the supermarket method began. It was originally company-confidential, distributed as blue-line copies of pencil-written pages. Each shop's process owner documented the production mechanism of his own area as a worked example.
かんばん事例集 (1970) — artoflean.com/reference/kanban
The casebook grew out of Hino tie-up trainees questioning Ohno
After a 1966 business tie-up with Hino Motors, roughly 400 Hino managers came to Toyota's head-office machinery plant for work training. At the close of training, discussion sessions were held with plant manager Taiichi Ohno, who explained the manufacturing thinking and methods built up over years. The basics distilled from dozens of these sessions, plus concrete shop examples, were assembled into the March 1970 casebook.
かんばん事例集 (1970), editor's account — artoflean.com/reference/kanban
The kanban method was first set in print in a Toyota Engineering Society piece
After the pencil-written 1970 casebook, the kanban method first appeared in typeset print when the Toyota Engineering Society (トヨタ技術会) published an explanation in its publication "Automotive Knowledge" (自動車の知識). The casebook also fed internal training: from around 1972 it was used as material for special foreman education, and the system was further developed into the "Toyota-style Production System — Basics" (January 1973) and "Applications" (September 1977).
Toyota Engineering Society, かんばん方式解説書; かんばん事例集 (1970) — artoflean.com/reference/kanban
The supplier-facing kanban was renamed the SD method by young engineers
When the internally established kanban method was extended to part procurement from suppliers, young university-educated engineers argued that "kanban" sounded too crude and named it the SD method, for Synchronised Delivery. Toyota's internal voucher registration for the card became the "SD card." The OB notes the irony that kanban later became an international word.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
One-piece flow inside Toyota is still constrained to small lots in many processes
The current-state reality across many processes, including inside Toyota, is small-lot production, not literal one-piece flow. Stamping lots run in the hundreds of parts because die changeovers, even after decades of setup reduction, require a minimum lot size. The direction is toward one-piece flow; the constraint is what physics and economics allow today.
Art Smalley, Art of Lean, Gemba Coach draft "Is One-Piece Flow Always Better?" — artoflean.com/reference/one-piece-flow
The Kamigo engine plant heat-treat lot was 40 parts
Art Smalley reports that the heat-treat lot size at Toyota's Kamigo engine plant was 40 parts when he worked there. Heating one part at a time in a furnace would waste enormous energy, so the lot size reflects a thermal and physical constraint, not a failure of TPS maturity. Casting and similar processes have comparable constraints, and signal kanban governs replenishment for these batch operations.
Art Smalley, Art of Lean, Gemba Coach draft "Is One-Piece Flow Always Better?" — artoflean.com/reference/one-piece-flow
Toyota press changeover fell from hours to single-digit minutes
Ohno states that Toyota's internal big-press changeover, which took two to three hours in the late 1940s, was cut to fifteen minutes by the late 1950s and ultimately to about three minutes by the late 1960s. The Toyota Engineering Society explanation describes press setup driven to three minutes and calls single-digit-minute setup "single-minute changeover" (シングル段取り). This setup reduction is what enabled small-lot, leveled production.
Ohno, Toyota Production System (1978); Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/smed
Sugimori found lead time proportional to setup time
The 1977 Sugimori paper reports that applying JIT in press shops was the hard case, and the authors concluded that lead time is proportional to setup time, with lot size following from it. They state that since 1971 production-engineering improvements cut setup on the 800-ton hood and fender press from about one hour to ten minutes, against four to six hours then typical in Western countries.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/smed
Sugimori's 1977 press data showed a one-day lot against a ten-day Western lot
The 1977 Sugimori press-plant comparison for hood and fender stamping reported Toyota at 0.2-hour setup with three setups a day and a one-day-use lot size, against a U.S. plant at six-hour setup, one setup a day, and a ten-day-use lot. Toyota ran 500 to 550 strokes per hour versus 300 at the U.S. plant. For low-demand parts under 1000 units a month, Toyota's lot ran as large as seven days of use.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977), Table 1 — artoflean.com/reference/smed
Toyota's 1977 assembly man-hours per vehicle were about half of Western plants
Sugimori's 1977 Table 2 reported the Takaoka plant completing a vehicle in 1.6 man-hours, against 3.8 in a U.S. plant, 4.7 in a Swedish plant, and 2.7 in a West German plant. The paper presented this as evidence that Toyota's labor productivity was the highest among the major-country auto industries at that time.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977), Table 2 — artoflean.com/reference/just-in-time
Toyota's working-asset turnover ratio far exceeded competitors by 1965
Sugimori's 1977 Table 3 reported Toyota's turnover ratio of working assets rising from 41 in 1960 to 66 in 1965 and 63 in 1970, against roughly 13 for a Japanese competitor and 5 to 8 for U.S. companies. The paper presented the very high turnover as a direct result of holding minimal inventory under JIT.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977), Table 3 — artoflean.com/reference/inventory
Bar-coded kanban let Toyota purge inflated card counts automatically
A Toyota OB describes how kanban counts crept upward over time, because adding a card when one seemed lost was easy but removing a surplus card had to be caught when it detached and was often missed. Since cards in circulation could not be counted directly, Toyota put a bar-coded serial number on each kanban and read empty cards as suppliers carried them back, building a system that ejected the inflated surplus. This "inflation kanban" purge is an example of IT used to enforce the card-reduction discipline.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Early kanban counts were computed by slide rule on a custom calculating disc
A Toyota OB recounts that kanban quantities were originally computed by hand on an abacus or a Tiger mechanical calculator, with daily required quantity and the gap between card-collection and supplier-delivery times as variables. As kanban use expanded, the calculation consumed too much time, so they transferred slide-rule scales onto two rotating discs; aligning daily requirement against delivery time read off the answer directly. This is a concrete instance of card count being a deliberate calculation, not a guess.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Toyota used a single-direction information system with interwriter teleprinters
The 1970 casebook and a Toyota OB describe production instructions to the assembly and unit lines being sent from a control room by an "interwriter," a phototelegraph-like device that reproduced pencil-written characters electrically on receiving paper tape. The OB explains that only five vehicles' worth of instruction was issued at a time; issuing no extra information avoided the need to correct it, applying information just-in-time. Control-room instructions were treated as plant-manager orders, and production contrary to them was not allowed.
かんばん事例集 (1970); Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/pull-system
Toyota controlled daily output to within plus or minus five vehicles of plan
A Toyota OB explains that daily final-assembly output was controlled to within plus or minus five vehicles of the plan, because variation at the final line amplifies into waste and strain upstream. If the line ran behind it worked overtime, which was authorized in ten-minute units, and if it ran ahead it stopped and cleaned until quitting time. For large disruptions, the shortfall was spread evenly across the remaining days of the month rather than recovered in a rush, because stable final-line output is what makes kanban production possible.
Toyota OB memoir, かんばん方式の裏話; かんばん事例集 (1970) — artoflean.com/reference/heijunka
The 1970 casebook describes computer-leveled assembly sequencing by car name, type, and function
The 1970 casebook documents the head-office assembly line leveling production in a defined order: balancing daily volume between two assembly lines, then by car name, then by car type, then by car model and function, always starting from the lowest-volume item. This leveling was performed on computers, including a system named COSMIC for conveyor scheduling and management information control. The casebook stresses that car models produced too few to run daily must be spaced at regular intervals, a condition often overlooked.
かんばん事例集 (1970) — artoflean.com/reference/heijunka
Toyota timed instruction issue by the longest lead-time point in the line
The 1970 casebook describes instruction timing set by the longest lead-time path. Vehicle-assembly instructions were issued about fifteen lots, roughly 2.5 hours, ahead of line-up, computed from trolley-conveyor loading ten lots ahead plus preparation work. Unit-assembly instructions were issued forty lots, about six hours, ahead, based on the longest path through a large-axle plant's brake withdrawal. Lots were three large or five small vehicles.
かんばん事例集 (1970) — artoflean.com/reference/lead-time
Ohno leveled the Corona to a fixed daily mix of body types
Ohno gives a worked leveling example: to make 10,000 Coronas over 20 working days, split into 5,000 sedans, 2,500 hardtops, and 2,500 wagons, means 250 sedans, 125 hardtops, and 125 wagons a day. On the line, a sedan is run every other unit and a hardtop and wagon every third unit, minimizing lot size and production variation. At the Tsutsumi plant, Corona and Carina alternated on one line rather than running all Corona in the morning and all Carina in the afternoon.
Ohno, Toyota Production System (1978) — artoflean.com/reference/heijunka
Toyota's monthly plan is communicated to suppliers as notice then confirmation
Ohno describes the Toyota information system as an annual plan, then a monthly production plan: for cars to be made in March, January gives a preliminary notice of what and how much, and February confirms the content down to model, type, and detail. The same notice and confirmation are sent to outside cooperators at the same times, so suppliers can review capacity and set kanban turnover counts. The Toyota Engineering Society explanation gives the parallel example of January's cars being notified in October and confirmed in mid-December.
Ohno, Toyota Production System (1978); Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/hoshin-kanri
Kanban governs goods, not orders, with no clerk involved in the order itself
A Toyota OB describes the 100-percent delivery method: a kanban detached from a box becomes, at the supplier's next delivery, the order for the next lot, with no clerk involved in placing it. To control whether the ordered amount was fully delivered, a delivery note and acceptance slip were issued with the order, and partial shipments were not allowed to be re-counted on the slip; the slip was held and displayed at receiving until 100 percent was completed. This created visual control of unfilled orders and simplified paperwork.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Toyota moved to no-inspection receiving under kanban
A Toyota OB explains that under kanban there is no spare inventory, so inspecting delivered parts and finding defects is too late and stops the line. Toyota therefore moved to acceptance without inspection, on the basis that the supplier guarantees quality and delivered quantity, starting with suppliers whose systems were ready. The OB believes the resulting awareness that stopping Toyota's line was serious drove suppliers to strengthen their own quality-assurance systems.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Toyota printed supplier logos on kanban to prevent sorting errors
A Toyota OB describes printing each supplier's company crest on the kanban, because an image is more reliable than text for preventing human sorting errors at a glance. Since computers of the era could not process images and cards were handmade, and crests were hard to draw by hand, the card paper was divided into three parts — supplier crest, part-information area, and receiving color band — each pre-printed separately and combined in a three-pocket kanban case. This is a concrete example of designing the card for error-proof handling.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/poka-yoke
Temporary kanban handled non-cyclic production such as holiday work
A Toyota OB explains that kanban can only connect processes operating cyclically, so mismatched weekly working days between Toyota and a supplier would break normal card rotation. For extra volume run on overtime or holidays, disposable temporary kanban were issued, split and released in small daily amounts in advance, and discarded after use. During the high-growth period the volume of temporary-kanban creation grew large enough that Toyota automated it with BASIC programs that printed characters and bar codes.
Toyota OB memoir, かんばん方式の裏話 — artoflean.com/reference/kanban
Kanban can manage irregularly used parts, not only steady demand
Ohno corrects the belief that kanban can only manage steadily-used parts, using the propeller-shaft balance weight. Five kinds of weight are chosen by the degree of rotational imbalance, used wholly irregularly; attaching a kanban to the physical weights made the kinds and quantities exactly confirmable, held the five stocks constant, and greatly reduced stock. The precondition is always knowing the exact stock of each kind at each process.
Ohno, Toyota Production System (1978) — artoflean.com/reference/kanban
Pull exposes instability rather than fixing it
Pull does not fix instability by itself; it exposes instability. Replenishment loops, sequential pull, and mixed systems all depend on demand pattern, process capability and yield, lead-time reliability, equipment uptime, changeover time, lot size, material availability, and management response when signals go abnormal. Installing pull on an unstable base surfaces shortages that no one then solves.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/pull-system
Reducing kanban count is the deliberate improvement mechanism
Kanban is not only a control tool; reducing the number of cards or container quantities can expose changeover, downtime, quality, material, layout, or response weaknesses. This must be done deliberately and with problem-solving support. Randomly cutting cards to reduce inventory is not kaizen; the correct response when a reduction causes disruption is to solve the underlying problem, not to add cards back.
Art Smalley, Art of Lean (JIT skill pack); Kanban encyclopedia entry — artoflean.com/reference/kanban
Kanban performs fine adjustment automatically within a range
Ohno calls kanban's great feature its automatic fine adjustment within a range, likening the system to an autonomic nervous system. Because no detailed plan is given to each line in advance, what model comes next is not known until a kanban comes off; if the expected mix reverses, workers simply make what the kanban say without anyone phoning around. The Toyota Engineering Society explanation similarly states that with the schedule chart removed, kanban absorbs production variation within a band automatically.
Ohno, Toyota Production System (1978); Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/kanban
Sugimori's 1977 paper states Toyota workshops no longer relied on a computer
The 1977 Sugimori paper states that with kanban, Toyota's workshops no longer relied on an electronic computer for scheduling. It gives three reasons: a fully computerized real-time scheduling system across all processes and suppliers, with constant alteration, is hugely expensive; using kanban itself lets managers perceive continuously changing facts without a computer; and kanban limits the surplus capacity that upstream processes are pushed to hold as demand grows more erratic further upstream.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/kanban
Overproduction is the worst waste and pull is its direct countermeasure
Sugimori states that under JIT the value of inventory's existence is disavowed; stock on hand is seen as a collection of troubles and bad causes, mostly the result of overproducing, the worst waste, because it hides the causes that should be fixed. The Toyota Engineering Society explanation says the four kanban rules prevent overproduction, which generates surplus workers, surplus equipment, and surplus inventory and is the most troublesome waste. Pull and kanban exist primarily to restrain this waste.
Sugimori et al. (1977); Toyota Engineering Society, かんばん方式解説書 — artoflean.com/reference/overproduction
Lead time is governed by inventory, so cutting WIP compresses it
Ohno drew a direct connection between lead time and the amount of work-in-process in the system; reducing WIP drops lead time proportionally. This is why Toyota's focus on eliminating inventory is about compressing lead time, not cost per se: every piece of inventory is time a product sits rather than flows. Sugimori frames inventory quantity and lead time as policy variables, which also exposes surplus equipment and workers.
Ohno, Toyota Production System (1978); Sugimori et al. (1977); Lead Time encyclopedia entry — artoflean.com/reference/lead-time
Late or missing kanban must trigger problem solving, not routine expediting
A core failure mode is late or missing cards becoming routine expediting instead of problem solving. Kanban should make abnormalities visible — late cards, missing containers, empty supermarkets, excess inventory, wrong quantities, or out-of-sequence production — and those abnormalities should trigger response, not workaround. If the organization cannot respond to the abnormalities pull exposes, the weakness is in management-system capability, not in low inventory itself.
Art Smalley, Art of Lean (JIT skill pack) — artoflean.com/reference/abnormality-management
Sugimori's stated ideal is an invisible total conveyor line
The 1977 Sugimori paper states Toyota's goal through kanban is a total conveyor-line production system linking all external and internal processes by invisible conveyor lines. When the policy factor, container capacity, and waiting time approach zero, one, and zero, the system has the attributes of a literal conveyor line: parts processed in flow, emerging one by one as completed vehicles, the whole line stopping on trouble and resuming immediately. Kanban is described as the conveyor connecting all the processes.
Sugimori, Kusunoki, Cho & Uchikawa, IJPR (1977) — artoflean.com/reference/just-in-time
Flow and pull are two of the basic JIT principles, alongside takt
Toyota's TPS glossary places downstream pull (後工程引き取り), continuous flow processing (工程の流れ化), and takt as the basic principles of just-in-time, with leveling (heijunka) as the premise. Flow without takt means flowing at a pace that suits the process rather than the customer; flow without pull means pushing material downstream regardless of need; flow without leveling means demand spikes propagate and destroy the flow. They operate as a system, not as independent choices.
Toyota TPS glossary, via Gemba Coach draft "Is One-Piece Flow Always Better?" — artoflean.com/reference/just-in-time
Sugiura ties leveled volumes to a president-level planning decision
Sugiura's Global Ten material shows production volumes as a president-level decision tied to domestic sales, export volume, and trade-friction considerations. It describes monthly production volumes of different models being leveled at assembly lines, with special handling for export vehicles that needed to be loaded on ships in time. This is a real planning system, not merely a shop-floor heijunka box.
Mikio Sugiura, Global Ten, via Art Smalley, Art of Lean — artoflean.com/reference/heijunka