Three hundred variants for one process: good news or bad?
The first analysis of a process almost always reveals hundreds of distinct paths. The count is not the problem: what matters is how much volume the first few cover.
A variant is a distinct sequence of activities: two cases going through the same steps in the same order belong to the same variant. On a moderately busy purchasing process, the first analysis routinely counts two to three hundred. The usual reaction is alarm. It is premature.
Read the curve, not the total
The number that matters is coverage: how many variants does it take to cover eighty percent of cases? If three are enough, the process is healthy and the tail is made of legitimate exceptions. If it takes sixty, there is no process: there are personal habits.
What the tail produces
Rare variants are not free. Each is an exception someone handled by hand, that no control caught, and that no procedure describes. They weigh on average lead time, they make forecasts wrong, and they are the leading source of disagreement between departments.
- A rare but long variant costs more than a rare and short one: always cross frequency with duration.
- A variant that exists only for one person points to a local practice, not a tooling defect.
- A variant that appears in a single month usually points to a configuration change everyone had forgotten.
Where to start
Do not aim to reduce the number of variants; that is an empty target. Take the three that cost the most cumulative time, trace their cause, and look at what the curve does next quarter. The rest can wait, and probably always will.
