Little's law
WIP law · queueing law · work in progress law
In a stable system, the average number of items in it equals the arrival rate times the average time each spends there. It means cycle time can be cut by reducing work in progress, without anyone working faster.
In practice
Twenty open projects at two completions a week means ten weeks from start to finish. Taking on fewer at once shortens every one of them, and nothing else has to change.
The common mistake
Assuming throughput rises when you start more work. Starting more raises work in progress and therefore lengthens the time each item takes, while completions stay at whatever the capacity actually is.
John Little proved it in 1961 and it holds for any stable system, regardless of how the work arrives or in what order it is handled. That generality is what makes it useful.
L = λW. The average number of items in the system equals the arrival rate multiplied by the average time each item spends in it. Rearranged, the form that matters operationally:
Cycle time = work in progress ÷ throughput.
What it says
You have twenty active client projects. You finish two a week. Then the average project takes ten weeks from start to finish, and no amount of urgency about any individual one changes that, because the arithmetic does not care about effort.
To shorten cycle time there are exactly two levers. Increase throughput, which means more capacity and costs money. Or reduce work in progress, which is free and is almost never done.
Cut to ten active projects at the same two a week and the average project now takes five weeks. Nobody worked faster. The same work completed in the same total time, and each individual item spent half as long in the system.
Why this is counterintuitive
Starting work feels like progress and an idle person feels like waste, so the instinct is to begin more things. That raises work in progress, which lengthens everything, while throughput stays at whatever capacity allows.
It also has a compounding cost, because items in progress are not free to hold: each one carries context-switching, status updates, and a client waiting and asking. High work in progress generates coordination load that reduces the throughput that was already the constraint.
Using it
Measure your three numbers. Count active items, count completions per week, divide. Most people are surprised by the result and it takes ten minutes.
Set a limit on work in progress and hold it. Nothing starts until something finishes. This is the whole mechanism of a kanban board and it is the cheapest operational improvement available to most businesses.
Resist starting work to look busy. Capacity is measured by what finishes, not by what is open.
It also identifies where to look when cycle time is bad, which is the same place the bottleneck points: throughput is set by the constraint, so adding work anywhere else only lengthens queues.
Concept web
Open the full webQuestions
What is Little's law?
That in a stable system the average number of items present equals the arrival rate times the average time each spends there. Rearranged, cycle time equals work in progress divided by throughput.
How does reducing work in progress shorten delivery times?
Because cycle time is work in progress divided by throughput. Halving the number of active items halves the average time each one takes, with no change in capacity and nobody working faster.
Why does starting more work not increase output?
Throughput is set by capacity, not by how much has been started. Starting more raises work in progress, which lengthens the time each item takes, and adds context-switching that can reduce throughput further.