← All tools

Flow playground

Little's Law made visceral. Sliders drive a living queue — watch waiting time explode as utilisation nears 100%, price your batch size, and find the lever that actually clears your backlog.

Core lesson

Read the queue first

Set the conditions, then use the diagnosis to decide which trade-off deserves a closer look.

Try:

The queue, live

Backlog 0 In progress 0 Done 0

What the simulation says

Export
Seeded simulation · ~2,000 simulated days · deterministic · the URL carries your settings

Optional lessons

Optional lessons from the same queue

Hold the queue conditions above constant, then isolate one trade-off at a time.

Optional lesson · backlog recovery

Triage your queue

There's a pile today and everyone has a favourite fix. This ranks the four standard levers — more people, less intake, smaller items, a WIP limit — against your queue: the sliders above set the world, plus one number.

Export
Same seeded simulation, one lever changed at a time

Why waiting dominates

Little's Law says items-in-progress = arrival rate × time-in-system. Push a team toward 100% busy and the queue — not the work — sets your delivery date. Near full utilisation, small bumps in demand or item size multiply waiting non-linearly (the queueing curve Reinertsen built The Principles of Product Development Flow around). That's why the verdict splits the average item's calendar time into working and waiting — the waiting is usually the bigger number, and it's the one your process choices control.

The WIP limit slider shows the other half of the argument: in-progress items share the team, so WIP above capacity doesn't add throughput — it just stretches every item's calendar time. "Start less, finish more" falls out of the arithmetic. But note what a WIP limit can't do: if demand exceeds capacity, the backlog grows without bound and no limit fixes that — the tool says so, loudly, rather than hiding the queue upstream.

The batch-size card prices Reinertsen's other classic trade with the same demand number: a release's fixed transaction cost argues for batching up, the cost of delay argues for shipping now, and the U-shaped sum of the two has a floor — your economic batch size. Triage ranks the four standard backlog levers — more people, less intake, smaller items, a WIP limit — against the queue you actually have. The expedite lane makes the hidden tax of priority visible, while dependent dice shows why locally reasonable capacity cannot promise system flow.

One honest caveat: this is a flow model of one stage. Real teams aren't a single pipe, and constraints in human systems are often policy, mindset, or coordination rather than a visible logjam — a lesson the Theory of Constraints crowd learned the hard way. Use this to win the argument about overloading; use judgement for everything else.