What is the Last Planner System?

The problem it was built for

On most jobs the programme is built once, at tender, by someone who will not be on site when the work happens. It is then issued downwards. Every trade is told when to start and when to finish. Almost nobody is asked whether they can.

So the programme describes what should happen. It never finds out what can happen until the week it fails. A trade turns up to an area that is not ready. A drawing answer arrives after the work needed it. Two gangs end up in the same room. None of these are planning failures at the moment they are noticed — they were decided weeks earlier, and nobody checked.

The Last Planner System was developed by Glenn Ballard and Greg Howell to close that gap. Its central move is almost embarrassingly simple: stop treating what should happen as though it were what will happen, and put two checks in between.

The four questions

Everything in the system hangs off four words. Traditional planning jumps straight from the first to the last and is surprised by the difference.

The gap between should and can is where most delay is created. The gap between will and did is where you learn what is really wrong with how you deliver.

The five levels of planning

The system is a cascade. Each level covers a shorter stretch of time in more detail than the one above it, and nothing enters a level until it has earned its place there.

What counts as a proper assignment

The weekly plan only works if the things on it are real commitments rather than wishes. Before a task is allowed on, it is tested against five questions.

Anything that fails a test goes back to the look-ahead to be made ready. It does not get put on the plan and hoped for. That single discipline is what protects the gang from a broken programme upstream.

Promises, not instructions

The part people underestimate is that this is a change in how people talk to each other. A traditional programme instructs: you start Monday. The Last Planner System asks for a promise, which means the person being asked is allowed to say no, or not yet, or not unless I get that drawing.

That sounds like a small thing. It is the whole mechanism. A date nobody agreed to tells you nothing when it is missed. A promise that was freely given, by someone who had the standing to decline, tells you a great deal — and people keep promises they chose to make far more reliably than dates they were handed.

Percent Plan Complete

The measure is deliberately crude: the number of promises finished divided by the number made, as a percentage, every week.

It is not a productivity measure and it is not an earned-value measure. It measures one thing — how much of what you said would happen actually happened. Teams new to it usually start somewhere around half. That number is not a failure; it is the first honest picture of how reliable the job really is, and most people find it uncomfortable precisely because it is accurate.

What makes it work is the second half: every missed promise gets a reason, the reasons get counted, and the ones that keep recurring get fixed. Asking why five times on a recurring reason usually walks you back from “the plasterers did not turn up” to something structural about how your business buys, informs or sequences work. That is the real output. The percentage is just what points you at it.

What it looks like when it is working

Nothing dramatic, which is the point. A short weekly meeting where trades commit to the week rather than being told it. A six-week view that people actually use to chase the things that would otherwise stop work. A daily fifteen minutes on site to keep the week honest. One number on the wall that everybody understands, and a list of reasons that gets shorter as the obvious problems get fixed.

The programme stops being a document that is defended in meetings and becomes something the people doing the work rely on, because they built it and they can trust it.