Sequential priority and WIP control

The fundamental building blocks of flow are, happily, mostly common sense. Start with what the T20 game proved: when we switch often, every project is delayed, capacity is lost, rework grows and stress rises. The round that worked was the one with stable priority: we decided what comes first and what comes next, and the priority didn’t change while work was in hand.

The fire exit

A meeting is under way in a hall. A fire breaks out, and there is one exit door. If everyone rushes for the door to get out as early as possible, the doorway jams and everyone is late, and some people may be hurt in the crush. The chance of getting everyone out safely is zero.

Is there another way? Form a queue. Decide who goes first, next, and next. Nobody pushes ahead. That guarantees the fastest evacuation with the least damage. A queue is a priority token system, and it is exactly what projects need: a sequential global priority.

But what about stuck tasks?

Stable priority alone isn’t enough. Imagine you are working on a task for the number 1 project and you get stuck: you need information from someone else, and it will take a while. So you pick up the number 2 project. That gets stuck too, while number 1 isn’t resolved yet. Do you open number 3? This can go on until you have many projects open again, and resolutions don’t arrive in sequence. You are back in bad multitasking.

So how many projects should be open?

The surgeon’s answer

Think of a renowned heart surgeon, always in demand. Their time is precious, so they aren’t present for the whole surgery: their team handles the preparation and the closing, and the surgeon steps in for the critical part.

If the surgeon operates on only one patient at a time, their precious time is wasted, and someone else who needed them goes without. If they start too many surgeries back to back, two patients may need them at the same moment, a matter of life and death.

They decide using four things they know:

1. How much of their time each surgery needs.
2. When in the surgery it is needed.
3. The natural waiting time: how long they are free in between.
4. How much protective capacity to keep aside for surprises.

From these, they decide how many surgeries to start, and stagger them so nobody’s life is put at risk. Try it:

The staggered sequence sets the stable priority. The number of surgeries started sets the work in progress. And the completion of one surgery decides when the next patient in the queue can start: a stable load in the system.

WIP and output

Plot the rate at which projects close against the number of open projects, and you get an inverted parabola. Too little WIP: resources idle, low output, lost capacity. Too much: the zone of bad multitasking, where output doesn’t just stop rising but falls sharply. In between is a plateau where output is at its maximum.

To move a system from the red zone to the green one:

1. Define a sequential global priority for all projects.
2. Reduce the number of open projects and control WIP.
3. Control the release of new projects.
4. Manage an active bucket and a waiting bucket perpetually to keep a stable load: one out, one in.

Setting the WIP limit in practice

A simple way to find the limit: rank all open projects, then assign resources starting from project 1 downward until every resource is covered. That number of projects is your starting WIP limit. Where natural waiting time is high, such as external approvals or customer inputs, the limit may need to be a little higher so people still have work while they wait, just as the surgeon uses waiting time between operations.

Reducing WIP at the start usually means freezing many open projects: they move to the waiting bucket and return one by one as active ones finish. It feels drastic, but those projects were barely moving anyway, and with the brake released, they start sooner than anyone expects.

Check your understanding

Answer all five. Get four right to complete the lesson.

1. The fire-exit story shows that…

• Everyone should rush
• A queue with a clear sequence gets everyone out fastest
• Exits should be wider
• Meetings are dangerous

Sequential priority beats everyone pushing at once.

2. Why isn’t stable priority alone enough?

• People ignore priorities
• Priorities change daily
• Projects are too small
• When tasks get stuck, people open more projects and drift back into multitasking

WIP control is needed as well.

3. Which is NOT one of the surgeon’s four parameters?

• How much of their time is needed
• When it is needed
• The patient’s age
• How much protective capacity to keep

The four are: how much, when, natural waiting time, protective capacity.

4. In the surgeon’s model, what decides when a new surgery can start?

• The completion of one already under way
• The surgeon’s mood
• The time of day
• The hospital’s revenue target

One out, one in: a stable load.

5. In environments with long natural waiting time, the WIP limit may need to be…

• Zero
• A little higher, so people have work while they wait
• Much lower
• Unlimited

Like the surgeon using the time between operations.