Every engagement opens the same way. One week, on site, following the work. No workshops in that week, no slides, no future-state diagram. We watch jobs move and we write down what happens to them.
Monday: pick the jobs
We choose three live jobs to follow, and we pick them with the people who run them rather than from a report. The aim is a spread: one that is going well, one that is stuck, and one that is ordinary. The ordinary one usually teaches us the most.
Tuesday and Wednesday: follow them
We sit with each desk the job passes through. We ask what arrived, what they did with it, what they needed that was missing, and where it went next. Then we ask the question that gives most of the answers away: why does this step exist?
Sometimes there is a good reason and it is on the tip of everyone's tongue. Sometimes the reason is a customer complaint from six years ago, handled by adding a check that has been running ever since.
Thursday: read it back
By Thursday we have a picture, and we walk it back through the same people in the same order. This is the part that matters. If the people doing the work do not recognise the description, the description is wrong, and we would rather find that out on Thursday than in a report.
Corrections at this stage are common and welcome. A step drawn in the wrong order, a handoff we missed, an exception that turns out to be half the volume.
Friday: write it down
The week ends with a short written summary: what the process is, where the work waits, and what we would look at first. Written, so it can be argued with. Three or four pages, in the same order the work happens.
What we do not do
We do not build a target operating model in week one. We do not benchmark anyone against a sector average. We do not ask people to fill in a time sheet describing their own work, because the answers are always tidier than the work.
A week is long enough to see how the work really moves, and short enough that nobody has time to tidy up for us.