Pull systems in lean manufacturing are production trigger mechanisms where no process produces anything until the process downstream of it signals that it needs material. The signal is the pull. Production flows in response to actual consumption rather than in anticipation of forecasted demand, eliminating the overproduction, inventory accumulation, and waiting waste that forecast-driven push scheduling generates.
Pull systems are one of the core operational tools of the Just-in-Time pillar of the Toyota Production System. JIT establishes the principle that each process should produce exactly what the next process needs, when it needs it, in exactly the quantity required. Pull systems are the mechanism through which that principle becomes operational on the production floor. Without a pull trigger, JIT remains a philosophy. With a functioning pull system, it becomes the daily production discipline.
This blog covers what pull systems are, why they work, the difference between pull and push, the three types of pull systems used in lean manufacturing, the tools that implement them, and the conditions a facility must establish before pull can function reliably.
Key Insight: Pull systems do not reduce inventory as their primary objective. They align production with actual customer consumption, and inventory reduction is the natural consequence of that alignment. Organizations that implement pull specifically to reduce inventory targets miss the systemic logic that makes pull reduction sustainable.
Pull Versus Push: The Fundamental Difference
Understanding pull requires understanding what it replaces. Most manufacturing facilities that have not implemented pull systems operate on a push model. Push and pull represent fundamentally different answers to the question of what should authorize production to occur.
How Push Production Works
In a push system, a central planning function, typically an MRP or ERP system, generates a production schedule based on demand forecasts. Each process in the production sequence receives a schedule telling it what to produce, in what quantity, and by when. The upstream process produces according to its schedule and pushes the output to the next stage regardless of whether that stage is ready to receive it or has capacity to process it.
Push production generates several predictable problems:
- Overproduction: upstream processes produce according to schedule even when downstream processes are not ready, accumulating inventory at every stage
- Inventory accumulation: work-in-process builds between every stage as a buffer against variability in both supply and demand
- Hidden problems: inventory buffers absorb and conceal quality problems, equipment unreliability, and process variability until they have compounded significantly
- Schedule complexity: central scheduling must coordinate every stage simultaneously, creating a planning burden that grows with production complexity
How Pull Production Works
In a pull system, production authority flows in the opposite direction. No process produces until the downstream process signals that it has consumed material and needs replenishment. The signal travels upstream. Production occurs in response to that signal. The downstream process pulls from the upstream process rather than receiving a push from a central schedule.
The result is a production system that is self-regulating at the operational level. When downstream consumption increases, more pull signals are generated and upstream production increases in response. When consumption decreases, fewer signals are generated and upstream production decreases. The system adjusts to actual demand without requiring a planner to update every schedule.
Pull systems also make problems immediately visible. Because there is no accumulating inventory to absorb disruptions, any breakdown, quality issue, or process deviation immediately affects the downstream process. The visibility this creates is not a vulnerability of pull. It is one of its primary benefits, and it is the mechanism through which pull and jidoka work together to drive continuous improvement.
Key Insight: Push production authorizes upstream processes to produce based on forecasts. Pull production authorizes upstream processes to produce based on actual downstream consumption. The difference determines whether inventory accumulates systematically or whether the production system self-regulates to actual demand.
The Three Types of Pull Systems
Lean manufacturing uses three distinct types of pull systems, each suited to different production environments and inventory management approaches. Understanding which type fits which context is essential for implementing pull correctly.
Type 1: Replenishment Pull
Replenishment pull, also called supermarket pull, maintains a defined inventory of finished or semi-finished goods between production stages. When the downstream process consumes units from the supermarket inventory, a replenishment signal is sent to the upstream process to produce exactly the quantity consumed.
The supermarket is not a warehouse. It is a controlled, defined inventory with a specific location for each item and a clearly defined maximum quantity. The maximum quantity is calculated to cover the replenishment lead time between the upstream and downstream processes. When the downstream process takes from the supermarket, the upstream process sees the kanban signal and produces to refill the supermarket to its defined level.
Replenishment pull is most appropriate when:
- Production lead time is longer than the delivery frequency the downstream process requires
- Product variety is high enough that sequential pull would create excessive changeover burden
- Process stability is sufficient to make replenishment quantities calculable
Type 2: Sequential Pull
Sequential pull, also called FIFO pull, produces units in the exact sequence and quantity that the downstream process requests. There is no supermarket inventory between stages. Each unit is produced specifically in response to a specific downstream request, and units flow through the sequence in the order they were requested.
Sequential pull is most appropriate when:
- Products are highly customized or have too many variants for a supermarket to hold all configurations
- Process lead times are short enough to produce to order within the downstream process's tolerance
- The production sequence can be tightly controlled and maintained
Sequential pull requires very high process stability. Any disruption to the production sequence, whether from quality, equipment, or material issues, directly affects the downstream process because there is no inventory buffer to absorb the disruption.
Type 3: Mixed Pull
Mixed pull combines replenishment and sequential pull for different product families or different stages of the same production sequence. High-volume standard products are managed through replenishment pull with supermarket inventory. Low-volume custom products are managed through sequential pull without supermarket inventory.
Mixed pull is the most common pull system configuration in facilities with significant product variety because it allows each product family to be managed with the pull type most appropriate to its volume and variability characteristics.
Key Insight: The three pull system types are not interchangeable choices. Replenishment pull suits high-variety, longer-lead-time environments. Sequential pull suits custom, short-lead-time environments. Mixed pull combines both for facilities with diverse product portfolios. Selecting the wrong type for the production context undermines pull system performance regardless of implementation quality.
The Tools That Implement Pull
Pull systems require specific tools to become operational. These tools make the pull signal visible, manageable, and self-regulating across the production sequence.
Kanban
Kanban is the most widely used pull signal tool in lean manufacturing. A kanban, meaning sign or signal in Japanese, is a physical or electronic authorization for upstream production or material movement. When downstream inventory reaches a defined replenishment point, the kanban signal releases, authorizing upstream to produce the consumed quantity.
Kanban exists in several forms:
- Production kanban: authorizes an upstream process to produce a specific quantity of a specific item
- Withdrawal kanban: authorizes movement of material from a supermarket to a downstream process
- Signal kanban: used with batch production processes, authorizing production when inventory falls below a trigger point rather than at each individual consumption event
- Electronic kanban (e-kanban): digital signals replacing physical cards, used in facilities with ERP integration or digital production management systems
The kanban quantity, the number of units each kanban card represents, is calculated from takt time, replenishment lead time, and a safety factor for process variability. This is why takt time stability directly affects kanban performance. When takt time changes, kanban quantities must be recalculated. When process variability is high, safety factors must be larger, which means more inventory in the system than a stable process would require.
Supermarket Systems
A supermarket in lean manufacturing is a controlled inventory point positioned between processes to supply a downstream process on demand. Unlike a warehouse, a supermarket has defined locations for each item, defined minimum and maximum quantities, and a direct connection to the pull signal system. When a downstream process takes from the supermarket, the withdrawal creates a visible replenishment need that the upstream process addresses.
The mizusumashi, or water strider, is the material handler who manages supermarket replenishment. The mizusumashi runs a defined route on a defined schedule, delivering material from supermarkets to production cells and returning empty containers or kanban cards to trigger replenishment. The mizusumashi separates material handling from value-adding production work, ensuring that operators in production cells are not spending production time retrieving material.
Point-of-Use Storage
Point-of-use storage positions materials and components at the workstation where they are consumed, eliminating retrieval trips that consume production time without adding value. In a pull system, point-of-use storage locations have defined quantities that trigger replenishment when they are depleted. The depletion is the pull signal.
Point-of-use storage is not simply proximity storage. It is a pull signal mechanism integrated into the physical layout of the production cell. The empty bin or the kanban square that has become vacant is the visible signal that the upstream supermarket or supplier needs to replenish.
Key Insight: Kanban, supermarket systems, and point-of-use storage work together as the physical infrastructure of a pull system. Kanban provides the signal. Supermarkets provide the controlled inventory from which downstream processes pull. Point-of-use storage makes the pull signal visible at the workstation level without requiring operators to leave their value-adding work.
What Pull Systems Require Before They Can Function
Pull systems fail consistently when they are implemented before the process conditions they require are established. The three conditions that must exist before pull can function reliably are the same three conditions that JIT requires: process stability, equipment reliability, and quality consistency.
Process Stability Through Standardized Work
Kanban quantities are calculated from takt time and replenishment lead time. Both calculations assume that process cycle times are reasonably stable. When cycle times vary significantly by operator or by shift, kanban quantities calculated from average cycle times will be wrong some of the time, generating either stockouts when the process runs slower than average or excess inventory when it runs faster.
Standardized work creates the cycle time consistency that makes kanban quantities calculable and reliable. Before implementing pull, every operation in the pull system's scope must have a documented standard that every operator follows consistently.
Equipment Reliability Through TPM
Pull systems have no inventory buffer between stages to absorb machine breakdowns. A breakdown in a pull system stops the downstream process immediately because there is no buffer stock to draw from while the upstream process is repaired. This is not a problem with pull. It is the mechanism through which pull makes equipment reliability consequential rather than absorb-able.
Total Productive Maintenance, through autonomous maintenance, planned maintenance, and OEE-guided improvement, creates the equipment reliability that allows pull systems to function without constant disruption. Implementing pull before establishing TPM disciplines creates a system that stops frequently not because pull is wrong but because the equipment is not reliable enough to support it.
Quality Consistency
Kanban signals authorize production of a specific quantity. When defective units are produced, they consume kanban authorizations without producing usable output. The downstream process receives fewer conforming units than the signal authorized, creating a gap in the flow. This gap either causes a downstream stockout or forces a manual override that bypasses the pull system entirely.
Quality consistency, achieved through standardized work, poka-yoke error proofing, and in-process quality checks, is required for kanban signals to authorize production reliably. The pull system and the jidoka quality system must be developed together, not sequentially.
Key Insight: Pull systems implemented before process stability, equipment reliability, and quality consistency are established generate chaos rather than flow. These three conditions are not prerequisites to be addressed before implementation begins. They are the ongoing disciplines that pull systems depend on continuously to function as designed.
Pull Systems Within the Lean System
Pull systems do not operate in isolation. They are the operational implementation of the JIT pillar and they depend on and interact with every other element of the lean system.
Pull and JIT
The relationship between pull systems and Just-in-Time is definitional. JIT is the principle. Pull systems are how that principle operates in practice. Takt time sets the pace. Pull systems trigger production at that pace from actual downstream consumption. Heijunka levels the demand variation that would otherwise make takt time unstable and kanban quantities perpetually miscalibrated. Value stream mapping reveals where the flow is disrupted and where pull systems should be implemented first.
Pull and Jidoka
Pull systems and jidoka are co-dependent in the same way that JIT and jidoka are co-dependent. Pull removes the inventory that would absorb quality problems. Jidoka ensures that quality problems exposed by pull are stopped and permanently resolved rather than worked around through manual overrides and buffer rebuilding.
Pull and the Three Enablers
Standardized work provides the cycle time consistency pull requires. TPM provides the equipment reliability pull depends on. Visual management makes the pull signal visible at every stage so anyone on the production floor can see whether the system is flowing as designed or whether an intervention is needed.
The kanban squares, supermarket locations, and point-of-use storage bins are all elements of visual management applied specifically to make the pull signal visible and self-managing.
Key Insight: Pull systems are the operational mechanism of the JIT pillar, implemented through kanban, supermarket systems, and point-of-use storage, sustained by standardized work and TPM, and connected to jidoka through the shared removal of the inventory buffers that would otherwise absorb and hide the problems both systems are designed to surface and resolve.
Q&A
Q: What is the difference between a pull system and a kanban system?
A: A pull system is the production philosophy and mechanism where upstream processes produce only in response to downstream consumption signals. Kanban is the most widely used tool for implementing pull, providing the physical or electronic signal that authorizes upstream production when downstream inventory reaches a replenishment point. All kanban systems are pull systems. Not all pull systems use kanban. Sequential pull systems, for example, use production orders tied to downstream requests rather than kanban cards as the pull signal.
Q: Why do pull systems fail in some manufacturing environments?
A: Pull systems fail most consistently when they are implemented before the three conditions they require are established. When cycle times are variable due to absent standardized work, kanban quantities are wrong and the system generates stockouts or excess. When equipment is unreliable, breakdowns stop downstream processes without buffer stock to absorb them. When quality is inconsistent, defective units consume pull authorizations without producing usable output. The pull system is not the cause of these failures. It is the mechanism that makes pre-existing instabilities immediately consequential rather than absorbable.
Q: How many kanban cards should a pull system have?
A: The number of kanban cards between two processes is calculated from three variables: the average daily consumption of the downstream process, the replenishment lead time from the upstream process, and a safety factor for process variability. The formula is: number of kanbans equals (daily consumption multiplied by replenishment lead time multiplied by safety factor) divided by the kanban quantity per card. As process stability improves and lead times shorten through lean improvements, the number of kanbans required decreases. Reducing kanban quantities over time is one of the primary mechanisms through which lean organizations drive continuous improvement in pull systems.
Q: Can pull systems work in high-mix low-volume manufacturing environments?
A: Yes, but the pull system type must match the production context. High-mix low-volume environments typically cannot use replenishment pull with supermarket inventory for every product variant because the inventory required to stock every variant would be excessive. Sequential pull, where each unit is produced specifically in response to a customer order, is more appropriate for custom and low-volume products. Mixed pull systems combine replenishment pull for the high-volume standard products and sequential pull for the low-volume custom products, giving each product family the pull mechanism most suited to its demand pattern.
Q: What is the difference between a supermarket in lean manufacturing and a warehouse?
A: A lean manufacturing supermarket is a controlled, defined inventory point with a specific location for each item, defined minimum and maximum quantities, and a direct connection to the pull signal system. Every item has a visual indicator showing its normal level and its replenishment trigger point. A warehouse is an uncontrolled storage location where inventory accumulates without defined limits, defined locations enforced by visual signals, or direct connection to a downstream consumption trigger. The supermarket is a tool of the pull system. The warehouse is the inventory accumulation that pull systems are designed to eliminate.
LeanSuite: A complete lean manufacturing software
Schedule Demo
