Five laws explain most of why work gets stuck between silos. They hold in a factory, an IT department, a kitchen and a hospital. Each one comes with everyday examples and a small model you can play with.
The law holds for any stable system, on average. It sounds obvious, but it has a hard consequence. If you can’t finish things faster, the only way to shorten lead time is to have less going on at once.
Want faster delivery? Stop starting, start finishing.
You walk into a café. Twelve people are ahead of you and the barista finishes about two drinks a minute.
You can estimate your wait before you join the line, without a stopwatch.
A team has 60 open tickets and closes about 10 a week. Every new ticket joins the back.
Adding a “quick” request doesn’t make it quick. It waits behind 59 others.
An office has 900 building permits in process and approves 150 a month.
Each case may only take a few hours of real work. The rest is waiting in piles.
Little’s Law works with averages. Kingman explains what happens when reality varies. The middle term is the trap: it isn’t a straight line. As utilization approaches 100%, waiting time shoots toward infinity, and more variation makes it happen sooner.
Planning people or machines for 100% utilization guarantees long queues. Short lead times need slack.
At 60% of capacity, one car braking hard changes nothing. At 95%, the same braking creates a queue that lasts an hour after the cause is gone.
A support team is planned at 100%. Then one urgent incident arrives. There is no slack to absorb it, so everything else slides, and the queue never quite recovers.
Managers are rewarded for keeping machines busy. They run large batches and start everything early. Utilization looks great. Lead times and inventory explode.
In The Mythical Man-Month, Fred Brooks argued that adding people to a late software project makes it later. Two things happen at once:
Don’t throw people at a bottleneck. Reduce dependencies and protect the team you have.
A project is three months late, so five consultants join. For the next six weeks the original team spends its days explaining the system. The delay grows.
A kitchen gets two extra cooks for a busy Saturday. They don’t know where anything is, they ask constantly and block the stove. Service gets slower.
A decision needs sign-off from six departments. That is 15 pairs who need to agree. Add two more departments and it becomes 28.
Goldratt’s The Goal introduced the Theory of Constraints. A system can never produce more than its narrowest step allows.
Find the bottleneck. Never let it stand still. Make everything else follow its pace. Then lift it.
In the book, a scout troop hikes in a line. The whole troop can only move as fast as Herbie, the slowest boy. Put him at the front and lighten his backpack, and everyone arrives sooner.
The airline adds more check-in desks. Passengers get through check-in faster and then stand longer at security. The trip to the gate takes exactly as long.
Every purchase over a limit needs sign-off from one manager who travels a lot. Faster forms and smarter systems do nothing while that inbox is the narrowest step.
Work expands to fill the time available for it. Parkinson wrote it as satire about bureaucracy, but anyone who has had a long deadline knows it’s true.
In flow terms, generous deadlines and big batches create hidden waiting. Short cycles, small pieces of work and a steady rhythm keep attention on what creates value.
Shrink the box: smaller pieces, shorter cycles, a fixed rhythm.
The calendar default is 60 minutes, so meetings take 60 minutes. Book 25 and most of them finish in 25.
Units that don’t use their budget lose it next year. So money that wasn’t needed in October gets spent in December.
Everyone adds safety margin to their task. Then each task starts late and uses its full margin, so the project is still late.
Smaller pieces of work vary less and wait less. They also give feedback sooner.
Set a cap on what is started. Queues shrink, lead time falls, and the bottleneck becomes visible.
Don’t plan people for 100%. Spare capacity is what absorbs variation and keeps lead times short.
A note on the curves: waiting time near full utilization isn’t technically exponential. It grows as 1 ÷ (1 − u), which goes to infinity at 100%. Communication channels in Brooks’s Law grow with the square of team size, not exponentially. Both are bad enough.