When nothing is wrong and nothing is moving

What it says

Five steps, and the order is the whole method. Find the one place work piles up. Get everything you can out of that place without spending anything. Make everything else serve it instead of competing with it. Only then spend money or people on making it wider. Then start again, because once it's wider the bottleneck has moved somewhere else. That last step is the one everybody skips. They fix the bottleneck, call it done, and are surprised six weeks later.

Where it comes from

Eliyahu Goldratt, in a novel from 1984 about a factory with three months to turn itself round. It's called The Goal. From there it spread through manufacturing and then into software and project work. Goldratt's own account was that he was putting intuition into words rather than measuring anything, so what sits behind it is worked examples rather than trials. Worth knowing before you quote it at somebody.

What to do with it

Take one piece of work and follow it from the first request to finished. Write down every place it sat still, and for how long. One of those waits will be much bigger than all the others. That's the constraint, and it's almost never the person who looks slowest — it's usually a step everything has to squeeze through. Then leave everything else alone until that step is better.

When it's the wrong lens

When it's stuck on a disagreement rather than a queue. Two people who haven't spoken have no bottleneck to widen, and this turns into a diagram of a conversation somebody needs to have.

What it says

In a system running steadily, the number of things on the go equals how fast you finish them multiplied by how long each one takes. Turn that round and you get the useful version: halve how much is open at once and you halve how long each one takes to get through. Nobody works faster. Nothing gets cut. The only thing that changed is how many things you've got in the air at the same time.

Where it comes from

John Little proved it in 1961, in the maths of queues. It's a proof rather than a claim about how people work, which is why it's on this page at all — it holds for the tills in a supermarket and it holds for work in a team. The catch is the word steadily. It wants a system roughly in balance, and a properly stuck project usually isn't one. So take the number as a strong hint rather than a law about your week.

What to do with it

Count what's open. Not the backlog — the things somebody has actually started and not finished. Put them on one list and count them. The number is usually the finding, and it's usually much bigger than anybody guessed. Then stop starting things. Finish three before you start one. It feels like doing less. It's how everything gets quicker.

When it's the wrong lens

When three things are open and all three are stuck anyway. Then the problem isn't how much is on the go, it's whatever each one is waiting on.

What it says

Describe a problem at full scale and it stops being something anybody can act on. It's too big to picture, so every possible first move looks pointless next to it, and doing nothing starts to look sensible. Weick's answer is to shrink what counts as a result. A small win is concrete, finished, and worth having on its own — modest, but real. String a few together and two things happen. You find out what actually works, and other people start wanting in, because something that's visibly moving attracts help in a way a plan never does.

Where it comes from

Karl Weick, 1984, working out why enormous social problems freeze the people trying to solve them. It's an argument from psychology rather than a study of projects. Weick is straight about the catch, too: this is opportunistic and untidy, because every win changes the situation and the next one turns up somewhere you didn't plan for. Which makes it a habit rather than a plan you can write down in advance.

What to do with it

Name one thing you could finish this week that would be worth having even if the rest never happened. That last bit is the test. "Write the spec" usually fails it. "Move one team across and leave the others where they are" usually passes. Then do it, and tell somebody. Telling people isn't showing off. It's the part that recruits the next person.

When it's the wrong lens

When the whole thing is stuck on one decision nobody's made. Then a run of small wins is a lot of movement around the problem, and the problem stays exactly where it was.

The move

One list, in one place: every piece of work somebody's started and not finished, with a name next to each one. Then count them, and say the number out loud.

Why it works

Nobody holds that number in their head, and everybody guesses low. The list tends to end the argument about whether the team's busy, because what it shows is everything half done rather than nothing started.

How it fails

Writing down the backlog instead. The backlog is everything nobody's started yet — enormous, and beside the point here. Only started-and-not-finished counts.

The move

Pick something that finished recently and follow it from the first request to done. Write down every point it waited, and roughly how long for. Days is precise enough.

Why it works

It splits waiting from working, and nearly all the time turns out to be waiting. It also moves the conversation off who's slow and onto where work sits still, which is both more accurate and much easier to have.

How it fails

Picking the one that went unusually badly. Pick a boring one — what you want is the normal path, not the worst week anybody remembers.

The move

Finish this sentence: "By Friday, ____ is done and worth having even if we stopped there." If you can't finish it, make the thing smaller until you can.

Why it works

It rules out the work that only makes sense as part of something bigger, which is most of what a stalled project is made of. And something finished and worth having is the only honest evidence that the thing can move at all.

How it fails

Naming a task instead of an outcome. "Review the designs" is a task. "The signup screen is agreed and nobody's reopening it" is a finished thing.