Process mapping
Ninety seven percent of that form's life is waiting
Value added analysis on one ordinary approval, step by step. The proportion is not unusual, and the part everyone tries to improve is the smallest number on the page.
Value added analysis has a reputation for being an exercise rather than a finding: a workshop, some sticky notes, a percentage nobody acts on. That reputation is mostly earned, and it comes from doing it on a map instead of on a real item.
So here it is on one real item. An expense approval, followed from submission to payment, with a stopwatch rather than an opinion.
The eight steps
Every step is timed twice: the time somebody was working on it, and the time from when it became available to when it left.
| Submitted and checked for completeness | 4 min work | |
| Waiting for the manager to open it | 31 h wait | |
| Manager reviews and approves | 3 min work | |
| Waiting in the finance queue | 52 h wait | |
| Finance checks against policy | 11 min work | |
| Waiting for the payment run | 46 h wait | |
| Paid | 2 min work | |
| Working 20 min, elapsed 129 h. Value added share | 0.26% |
Illustrative figures from a composite of ordinary approval flows. Twenty minutes of work spread across five and a half days. Nothing in this claim went wrong and nobody involved was slow.
Twenty minutes of work in a hundred and twenty nine hours of elapsed time. Not because anybody was slow: every individual step was done promptly once it was picked up. The time is entirely in the four gaps between people.
The number is not the point
A quarter of one percent sounds like a scandal and it is not. Ratios like this are completely normal for any process that hands work between people, and quoting the percentage on its own is what makes the exercise feel like theater. Nobody can act on 0.26 percent.
What is actionable is which gap is biggest and why it exists.
| Gap | Hours | Why it is there |
|---|---|---|
| Finance queue | 52 | Batched. Claims are looked at when somebody sits down to a batch of them |
| Payment run | 46 | Scheduled. Runs twice a week whatever is waiting |
| Manager’s inbox | 31 | Unbatched, but competing with everything else in an inbox |
Two of those three are policy decisions rather than performance problems. The payment run is twice a week because somebody decided twice a week. That is a defensible decision and it may well be the right one, but it is currently costing the process forty six hours per claim and nobody is pricing it.
What the exercise usually gets wrong
It improves the twenty minutes. Faced with this table, the standard response is to look at the eleven minute finance check, because that is the biggest work number and it is the one somebody feels responsible for. Halving it saves five minutes out of a hundred and twenty nine hours. It is not measurable in the outcome.
It counts the wrong thing as waste. The four minute completeness check adds no value to the claimant and it prevents rework downstream. Deleting it makes the process worse. Work that is not value adding is not automatically removable, and the distinction between waste and necessary non-value work is the whole craft of this technique.
It stops at the percentage. The percentage is the headline. The gap table underneath is the deliverable.
What to do with it
Follow one real item, end to end, with two clocks. Work time and elapsed time. One item is enough to be shocking and cheap enough that it actually gets done, which a study of forty never is.
Rank the gaps, not the steps. Then for each gap ask which of three things it is: batched, scheduled, or queued. Each has a different fix and only the third is about capacity.
And price the largest one before defending it. Twice a week may be exactly right for the payment run. But the decision costs forty six hours a claim, and that belongs on the table when it is made rather than inherited.
The finding is never the percentage. It is that the three biggest numbers in the process were nobody’s job to look at.