What a process audit really finds (with an example)

The symptom

Every business has a favourite explanation for its problems. "We need a second CNC machine." "Assembly is too slow." "Sales keeps promising dates that production can never hit." These lines come out in the first half hour of any conversation, delivered with great confidence.

What's striking is this: in the majority of cases I have seen, the named cause was not the real one. Not because the people are incompetent — quite the opposite, they know their business inside out. It's because a bottleneck almost always shows up where the work backs up, and only rarely where it originates. You see the queue, not the source.

The mechanism

The reason is simple, and a motorway traffic jam illustrates it: you are stopped at kilometre 50, but the cause — roadworks — is at kilometre 55. Where you experience the standstill is not where it is caused. The same holds on the shop floor. A full buffer in front of assembly looks like an assembly problem. In reality, assembly can be perfectly healthy and simply catching everything that goes wrong two steps upstream.

On top of that comes a perception effect: you remember the loud moments. The machine that broke down last week burns itself into memory. The quiet losses — the order that waits two hours for a sign-off every single day — go unnoticed, because they have become normal. Yet that invisible, chronic waiting is usually the bigger item. An audit therefore measures the whole flow instead of chasing the loudest complaints.

And there is a third effect I run into almost everywhere: every department optimises for itself. Production reports full utilisation, sales reports full order books, work scheduling reports that it can barely keep up. Each of those statements is true on its own — and yet none of them tells you where the overall process is stuck. The truth lives at the handoffs between departments, and no one feels solely responsible for those. That is exactly where an audit looks first.

The cost

An illustrative example. The business and the figures are invented. Take a maker of metal assemblies, a good 80 employees, that is certain: the bottleneck is machining. The machines run at the edge of capacity, and orders pile up in front of them. The obvious plan: an additional machine for around 220,000 euros, plus a second shift.

An audit first measures the path of an order from receipt to dispatch, with timestamps and not with opinions. In the example it shows that machining is not the bottleneck. Orders back up before it because the technical sign-off of the production paperwork takes eleven working days on average. Drawings wait for a check that a single person handles on the side. Only once that step releases a batch of orders do they hit the machines all at once, and then it looks like a machine problem.

Let's run the false diagnosis through the numbers: 220,000 euros of investment plus the running cost of a second shift, roughly 120,000 euros a year. The extra machine would not touch the sign-off backlog. The orders would still sit eleven days upstream. The money would go into capacity that never touches the actual problem.

The fix

The real bottleneck can be relieved without any investment. Sign-off moves from one person to a small two-person team, part of the checking becomes a checklist others can work through, and the sequence follows delivery date rather than order of arrival. The example assumes that sign-off time then falls from eleven working days to two. The buffer in front of the machines melts away, and the existing CNC machines are enough.

A result like that also changes the conversation on the floor. Machine operators who have felt for months that they are too slow see something else in the numbers. That side effect of a clean audit is easy to underestimate: it takes the blame off the people who have been carrying it unfairly, and turns attention to the process instead of the individuals. When you measure instead of guess, the conversation gets calmer, too.

The point is not that every business has its problem in sign-off. The point is the method: measure first, then decide. Don't add capacity where the queue shows up; find the source where it originates. That costs a few days of clean observation, and in this example it avoids a six-figure misinvestment.

That is exactly the heart of a process analysis: map one of your core processes end to end, measure with real timestamps where time and money are lost, and deliver a prioritised list of actions — before anyone decides on an investment. Fixed scope, fixed duration, fixed price.

The next step

If you have a larger investment coming up to solve a visible bottleneck, it can pay off to spend a few days measuring first — checking whether the bottleneck really sits where it shows.