How the work actually runs.

Work is not lost inside a stage, it is lost between them. These are the five joints every operation I have run has had, what goes wrong at each one, and the smallest control that stops it. Pick a joint to see what the control costs, because a control is never free and that is the part nobody writes down.

The controls total less than a day a week. An operation that needs more control than that is not under-managed, it is badly designed.

I use these models every working day and I still think most of what is claimed for them is nonsense. So here is the judgement written down instead of argued: every task placed by how much judgement it really needs and what a mistake actually costs. Below the line I hand it over without thinking. Above it I do not, and no demo has moved my mind yet.

The line is not about capability, it is about who answers for the mistake. That is the whole argument and it does not move when the models get better.

Written out because every hiring conversation asks it and most answers are improvised. This is the one I would give, and it is deliberately unexciting for the first two weeks.

1. 01
   Days one to fourteen

   Watch, and change nothing

   I sit in the actual work and write down where it stalls. No opinions in week one, because the thing that looks stupid from outside is usually load bearing. By day fourteen I can draw your operating picture on one page and name the two joints that are costing you the most.

   • Sit through the real process end to end, twice
   • Write the operating picture on one page
   • Find who actually decides, as opposed to who is listed
   • Name the two joints losing the most work
2. 02
   Days fifteen to forty five

   Fix one joint properly

   One joint, the worst one, fixed to the point that it stays fixed after I stop looking at it. Not a programme. A control small enough that the team keeps doing it in month six, which is the only test that counts.

   • Fit the smallest control that holds
   • Write it down in the words the team already uses
   • Agree the one number that says whether it worked
   • Hand it to a named owner, not to a process document
3. 03
   Days forty six to ninety

   Prove it, then do it again

   Read the number. If it moved, take the second joint and repeat. If it did not, say so plainly and work out why rather than adding more process on top of process that is not working. By day ninety you have two joints holding and a written list of what I would do next, in order, with what each one costs.

   • Read the number honestly, including when it is bad
   • Take the second joint
   • Write the ordered list of what comes next
   • Make sure it all survives without me

If the first ninety days produce a reorganisation chart, something has gone wrong. They should produce two or three things that work and a clear list of what does not.

This is the actual job, as a game. You have one quarter, forty hours a week, and five places to put them. Everything trades against something else, the weeks do not wait for you, and the quarter ends whether or not you are ready. Play it once and you will know more about how a practice like this really runs than any case study on this page can tell you.