# Little's Law and WIP Limits

Explore how work in progress shapes lead time and compare two deterministic flow simulations.

Duration: 12 min

When throughput is steady, more work in progress means more waiting. A deterministic simulation lets you compare a crowded system with a constrained one without treating the model as a production forecast.

## fmap-m4-l2-v1

The approved narration and storyboard are present in the source pack. The text digest and deterministic simulator provide reviewable non-video paths.

## fmap-m4-l2-digest

### Little's Law and WIP limits

For a stable system, Little's Law gives a useful relationship: **average lead time = average work in progress ÷ average throughput**. It is a long-run relationship, not a promise about one item.

Imagine a coffee shop with twenty customers inside and a steady throughput of ten customers served each hour. The average experience is two hours. If the same baristas keep the same throughput while only ten customers are admitted into the system, the average falls toward one hour. More started work did not create more finishing capacity; it created waiting.

A WIP limit is an agreement not to start another item when a stage is full. It can feel like imposed idleness, but the visible spare capacity is a signal to help finish blocked or queued work. “Stop starting; start finishing” is a system rule, not a demand for individuals to remain busy.

The relationship depends on a reasonably stable boundary and period. Arrival spikes, changing capacity, different item sizes, and blocked work all affect observed results. Use the law to reason about direction and system design, then inspect real data rather than treating the equation as a precise forecast.

Admitting fewer customers can move the queue outside the shop. The shorter average inside does not establish a shorter end-to-end wait or greater capacity. Measure waiting before admission as well, and preserve an accessible, safe route for people who cannot wait. A WIP limit should improve the whole service rather than hide unfinished demand beyond its counting boundary.

The accompanying simulation compares the same work and capacity under two WIP conditions. It is deterministic so the learning path is reproducible; it is not a production model.

## fmap-m4-l2-i1

This deterministic teaching model makes the Review constraint repeatable for keyboard, pointer, and text-fallback paths. It demonstrates direction; it is not a production forecast.