Manufacturing & OperationsGlossary

What Is Kanban?

Also known as: pull system, kanban card

Definition

Kanban is a visual pull system in which a card, bin, or electronic signal authorizes production or replenishment of a specific part only after downstream consumption, capping inventory at the number of signals in circulation.

Kanban Explained

The mechanics are simple and strict. A defined quantity of a part lives in a container with an attached signal. When the consuming operation empties the container, the signal is returned to the supplying operation or vendor, authorizing exactly one replacement quantity. No signal means no production, which is the essential difference from a push system where a schedule authorizes work regardless of downstream need. Total inventory for that part can never exceed the number of signals multiplied by the container quantity.

Sizing determines whether kanban works. The standard calculation is number of cards equal to average demand during replenishment lead time, plus a safety factor, divided by container quantity. If a cell consumes 200 pieces per day, replenishment takes 1.5 days, containers hold 100 pieces, and a 20 percent buffer is applied, the system needs (200 x 1.5 x 1.2) / 100 = 3.6, rounded up to four cards. Undersize and the line starves; oversize and kanban degenerates into an expensive way to hold the same inventory as before.

Kanban depends on demand stability and part repetitiveness, which is why it belongs on runners and repeaters rather than strangers. A useful screen is Pareto by usage frequency: parts consumed most weeks with moderate variability are good candidates, while parts ordered twice a year should stay on MRP. Attempting to kanban an entire bill of material is a common failure mode that produces thousands of stagnant cards and a loss of organizational confidence in pull systems generally.

Modern implementations are usually electronic. ERP systems including Infor SyteLine and LN support kanban or replenishment-signal functionality where a scan of an empty container generates a replenishment order or a supplier release, and consumption is backflushed rather than transacted operation by operation. The critical integration question is how kanban-controlled items are treated by MRP: if MRP continues to plan them independently, the plant ends up with two competing authorization systems and duplicate orders.

Why It Matters

  • Caps inventory by design rather than by policy, because physical signals impose a hard ceiling that cannot drift upward unnoticed.
  • Removes the need to schedule every internal operation, cutting planner workload and eliminating a large class of expediting.
  • Makes abnormal conditions visible immediately, since a missing card or an aging container is obvious on the floor without a report.
  • Shortens replenishment cycles for high-runner parts, freeing planners to focus attention on genuinely variable, low-volume demand.

In Practice

A plant kanbans 340 fastener and hardware SKUs into two-bin racks at point of use, sized on 30-day usage history. Six months later a quarter of the bins have not turned. Investigation shows those parts were carried over from a discontinued product family, and nobody owned card retirement. The fix is a standing quarterly review that recalculates card counts from current usage and physically removes cards for dead parts. Without that loop, every kanban system silently inflates back toward the inventory level it replaced.

Frequently Asked Questions

Can kanban and MRP coexist in the same ERP?

Yes, and most plants run both. The requirement is a clean split: kanban-controlled items must be excluded from normal MRP order generation, usually through a planning code or replenishment method flag, so only the kanban signal creates supply. Running both planners on the same part produces duplicate orders and inventory growth that is difficult to trace.

Which parts are wrong for kanban?

Parts with lumpy or seasonal demand, engineered-to-order items, expensive low-usage components, parts with long or unreliable supplier lead times, and anything approaching end of life. Kanban assumes consumption in the recent past predicts consumption in the near future. Where that assumption fails, forecast-driven MRP planning with explicit safety stock is the more appropriate mechanism.

Working with Kanban in a live environment? Our engineers do this every day - and our AI agents automate most of it.