An explicit cap between operations. The rule that makes pull actually pull.
WIP limits are the most underappreciated piece of pull system design. Most lean explanations cover kanban, supermarkets, and takt time in detail and then mention WIP limits in passing, as if they happen automatically. They do not. Without explicit caps, enforced physically and culturally, what looks like a pull system silently becomes a tidy push system with cards. The cap is what forces pull to actually pull.
"If there is no cap, there is no pull. The cards become decoration."
A WIP limit is a hard ceiling on how much in-progress material can sit in a specific place. The place might be a buffer between two operations, a queue in front of a single station, a kitting area waiting for assembly, or any other location where work piles up. The cap is set deliberately, based on how much material the team wants to allow in transit there, and it is enforced as a rule: when the cap is hit, the upstream activity stops.
The cap usually starts from standard WIP math. If two operations have cycle times of three and four minutes, and the team wants no more than ten minutes of work sitting between them, the cap is roughly two and a half parts, rounded to two or three depending on how the team wants to err. The math gets refined over time. The starting point is just a defensible number.
Enforcement is where most shops fail. A cap that lives on a screen, or in a policy document, gets ignored within a month. A cap that lives as marked floor space, sized exactly to the cap, gets respected because there is no room for a fourth cart when only three rectangles are painted. The physical constraint does the enforcement automatically. The team does not have to be disciplined; the floor is.
When the cap is hit, the upstream operation stops. That stop is uncomfortable the first few times it happens. The operator feels idle. The supervisor feels production is being lost. Lean shops train through this discomfort because the stop is the system working as designed. The stop is what surfaces the real problem: the downstream operation is slow, or the takt is wrong, or the schedule has been overloaded. Without the stop, none of those problems gets named, and the value stream keeps quietly choking under growing WIP.
Picture a small machine shop running CNC parts through five operations: turn, mill, deburr, inspect, pack. Without WIP limits, the turner runs full bore and stacks parts in front of the mill. The mill runs full bore and stacks parts in front of the deburr. Within a week, there are 200 parts in various states scattered across the shop, and the inspector cannot keep up with what is reaching them. The owner thinks the inspector is the bottleneck and considers hiring another.
The owner installs WIP limits instead. Between turn and mill, three carts maximum. Between mill and deburr, two carts. Between deburr and inspect, one cart. The floor gets painted accordingly. When the three rectangles between turn and mill are full, the turner stops and helps the mill or runs setup for the next job. Within two weeks, the team learns where the actual bottleneck is (deburr, not inspect, because deburr was being starved by inconsistent batch sizes from mill). The bottleneck gets addressed with a small process change. Total WIP across the shop drops from 200 parts to about 30. Lead time drops by a third. The owner does not hire another inspector.
A WIP limit is the enforcement mechanism for work in process sizing, often anchored to standard WIP. It is the rule that makes kanban actually create pull instead of decorated push. WIP limits are also a defining feature of any real pull system; without them, a system that calls itself pull is producing whenever the upstream station feels like it. Tight WIP limits are what compress lead time by forcing queue time out of the value stream.
The questions we hear most about this term.
Long-form guides that pick up where this definition leaves off, written for manufacturers running Arda today.
Same-day setup. No distributor lock-in. Zero stockouts. Top teams double revenue in 9 months.