Kanban card calculation sizes a pull system: it sets how many cards, and therefore how many containers of parts, circulate between a process and the supplier that feeds it. The standard formula is N = (D × L × (1 + S)) / C: average daily demand times replenishment lead time, inflated by a safety factor, divided by container quantity.
That one line is the whole job, but each input hides a decision that changes the answer. This guide works the formula through with real numbers, shows where each input comes from, and covers the part most kanban failures trace back to: sizing the loop once and never touching it again. If you are new to pull systems, start with our overview of kanban in manufacturing then come back here to size one.
What is the kanban card calculation formula?
The kanban card calculation formula is N = (D × L × (1 + S)) / C, where N is the number of kanban cards, D is average daily demand, L is replenishment lead time in days, S is the safety factor as a decimal, and C is the container quantity. It tells you how many cards to put in circulation so the downstream process never starves and the upstream process never overproduces.
The logic is simpler than the algebra. D × L is the demand that occurs during the time it takes to refill an empty container, the parts you consume while you wait. Multiply by (1 + S) to add a cushion for demand spikes and lead-time variation. Divide by C because each card represents one container, not one part. Round up: a partial card does not exist on the floor.
Where does each input come from?
Each input is a real measurement, not a guess, and getting them honest matters more than the arithmetic. Here is where to find each one.
Daily demand (D) is the average number of parts the downstream process consumes per day, taken over a representative window, usually the trailing four to twelve weeks. Use the demand at the point of consumption, not the sales forecast. If demand is seasonal or lumpy, size for a busy-but-normal period, not the single worst day and not the annual average.
Lead time (L) is the total elapsed time to refill one empty container, from the moment a card signals a need to the moment the full container is back and available. It includes the waiting, the setup, the run time, the transport, and any queue in between, not just machine cycle time. Lead time is where most people undercount, because the queue and transport are invisible in a cycle-time study. Measure the real thing.
Safety factor (S) is the cushion for demand spikes and lead-time variability, expressed as a decimal, 0.1 is a 10% buffer, 0.5 is 50%. Toyota's own guidance is conservative: keep the safety factor small, on the order of 10%, and attack the variation instead of padding around it. A big safety factor hides problems; it does not solve them. Start low, and only raise it when a documented instability forces you to.
Container quantity (C) is how many parts fit in one standard container. Smaller containers mean more cards, smoother flow, and faster feedback when something breaks, at the cost of more handling. Larger containers cut handling but coarsen the signal. Toyota's rule of thumb is to keep a container to no more than about 10% of daily demand so the loop stays responsive.
How do you calculate kanban cards with real numbers?
Work a real example and the formula stops being abstract. Say a downstream assembly cell consumes a bracket at an average of 600 parts per day. The feeder cell that makes the bracket takes 1.5 days to replenish an empty container once it gets the signal, run time plus the queue in front of the press. Each container holds 100 brackets. You set the safety factor at 0.2 (20%) because demand runs a little spiky.
Plug it in: N = (600 × 1.5 × 1.2) / 100 = 1080 / 100 = 10.8, which rounds up to 11 cards. Eleven containers of 100-1,100 brackets, circulate in the loop. That is the maximum inventory the pull system will ever hold between these two cells, and it is now a visible, capped number instead of whatever piled up.
| Input | Symbol | Value | Where it came from |
|---|---|---|---|
| Daily demand | D | 600 parts/day | Trailing 8-week average at the assembly cell |
| Lead time | L | 1.5 days | Card-to-full-container, including queue |
| Safety factor | S | 0.2 (20%) | Judgment: demand runs spiky |
| Container quantity | C | 100 parts | Standard tote size |
| Result | N | 11 cards | 10.8 rounded up |
Notice what the safety factor did. At S = 0, the loop needs 9 cards. At S = 0.2, it needs 11. At S = 0.5, it needs 14, half again as much inventory sitting on the floor. That is why Toyota pushes the factor down and the variation out: every point of safety factor is inventory you pay to hold.
How does a kanban loop actually circulate?
A kanban card is a recycling signal: it travels with a container, comes back empty, and authorizes exactly one refill. When the assembly cell empties a container, the card on it goes back to the feeder cell as a build order. The feeder makes one container's worth, attaches the card, and sends it forward. Nothing gets built without a card, so the number of cards is a hard ceiling on inventory. That is the point, the calculation is really setting your maximum work-in-process, not just a card count.
What are the steps to size a kanban loop?
Sizing a loop is a short, repeatable procedure. Run it the same way every time and the number defends itself.
- Pick one part and one loop. Size one supplying-to-using relationship at a time. Do not try to size the whole plant at once; each loop has its own demand and lead time.
- Measure daily demand (D). Pull consumption at the point of use over a representative window, and use the average per working day. Note whether demand is stable or lumpy.
- Measure the full lead time (L). Time a real replenishment end to end, card returns, part gets made, container comes back and is available. Include queue and transport, not just run time.
- Set the container quantity (C). Choose the smallest standard container that handling can support. Keep it near or below 10% of daily demand so the signal stays responsive.
- Choose a safety factor (S). Start small, 0.1 to 0.2, unless a documented instability justifies more. Write down why you picked it.
- Compute N and round up. N = (D × L × (1 + S)) / C. Round to the next whole card and put that many in circulation. Nothing else may be built.
- Run it, then resize on change. Watch for starvation (too few cards) or overflow and stagnation (too many). Recalculate when demand, lead time, or container size actually moves, not on a whim.
Kanban card calculation: by the numbers
The formula and its guardrails are documented in primary lean references, not folklore:
- The kanban concept and its pull logic come from the Toyota Production System where kanban is the tool that operates just-in-time production and caps overproduction.
- The Lean Enterprise Institute defines a kanban as a signaling device that gives authorization and instructions for the production or withdrawal of items in a pull system, the card whose count you are calculating.
- Toyota's practical guardrails, small safety factors on the order of 10% and container sizes near 10% of daily demand, reflect the system's bias toward exposing variation rather than padding around it.
When should you resize the loop?
Resize when a real input changes, not on a calendar. The number of cards is only correct for the demand, lead time, and container size you measured. When any of those moves, the loop is either starving the line or hoarding inventory, and the fix is a recalculation, not a workaround.
The common triggers: demand shifts up or down by more than a container or two per day; lead time changes because you added a shift, moved the feeder, or reduced setup with standard work; or you change container size. A quieter trigger is improvement itself, when you cut lead time in half, the loop should shed cards to match, and if it does not, you have just locked in the inventory you worked to remove. Resizing is where kanban stays honest. Treat the card count as a living number tied to a value stream you actually watch, the same way leader standard work keeps the daily routine current. And keep the loop visible, a board that shows cards in play is the simplest form of visual management there is.
The hard part in most plants is not the arithmetic; it is knowing your real demand and lead time in the first place. When consumption and replenishment are logged on paper, the inputs to this formula are stale before you finish adding them up, and the loop drifts out of size unnoticed. When capture is digital and current, resizing becomes a five-minute check instead of a data-archaeology project, which is the shift CLS made when it moved production data off clipboards and into a live feed. Size the loop with honest numbers, keep the numbers honest, and kanban does exactly what it promises: caps inventory, exposes problems, and never starves the line.