Throughput: how much the system really delivers
Delivering more is not starting more. Throughput is capped by capacity. Starting in excess only lengthens the queue. Adjust the WIP below, run the simulation and watch the delivery curve plateau.
A team gets steady demand and can focus on ~4 items at once. Throughput is how much it actually delivers per week. Raise the WIP and find out: is delivering more just a matter of starting more?
What throughput is
Throughput is how many items the system completes per period: per week, per sprint, per month. It is an output metric, measured at the end of the flow: what matters is not how much work was started, but how much made it all the way to delivery.
Because it counts only what finishes, throughput is honest. It ignores the effort stuck mid-flow and reveals the team’s real delivery capacity: the basis for forecasting and for measuring flow improvement.
Why starting more does not raise throughput
Intuition says starting more items delivers more. But throughput is capped by the system’s capacity (its bottleneck), not by how much you start. Once capacity is busy, every new item simply joins the queue.
This is the throughput plateau: the delivery curve rises, hits the capacity ceiling and flattens. Extra work does not become faster delivery; it becomes waiting. Worse: too much WIP adds context-switching, which burns capacity and can even drag throughput down.
Little's Law and the ceiling of the system
WIP ≈ Throughput × Lead Time
With throughput at the capacity ceiling, Little’s Law shows the price of pushing more work in: raising WIP raises lead time proportionally. To truly deliver more, the lever is not to start more: it is to expand capacity or relieve the bottleneck.
Frequently asked questions about throughput
What is throughput in Kanban?
It is how many items the system completes per period (week, sprint, month). An output metric, measured only on what crosses the flow to delivery.
What is the difference between throughput and velocity?
Throughput counts delivered items, regardless of size. Velocity sums estimate points. Throughput is simpler and needs no estimation: you just count what got done.
Does starting more items raise throughput?
No. Throughput is capped by capacity, not by how much you start. Past the ceiling, starting more only thickens the queue and raises lead time: the throughput plateau.
How do you actually raise throughput?
By acting on capacity: relieve the bottleneck, cut rework and interruptions, or add capacity where it is missing. Pushing more work in does not work.
Enjoyed learning by doing?
Take flow to the next level: manage a squad in the Kanban Game.
