Operations

Which improvement project should come first?

Mohit Saini · 19 Sept 2023 · 5 min read

Which improvement project should come first?

Every plant I visit has a list of improvement projects: set-up reduction, layout changes, 5S, Six Sigma, automation, quality circles. Each has a sponsor, a target and a review meeting. Most plant heads are proud of how many they are running.

I usually ask a different question: how do you decide which improvement comes first, and which shouldn’t be done at all?

The answer matters more than most managers realise.

Two plants, one lesson

Picture two plants of the same company, making the same range of products at about the same volume.

The first plant has been through a Lean programme. It is clean, well laid out and pleasant to walk through. The second is cluttered, with work in progress stacked in every aisle. The manager is understandably proud of the first and plans to transform the second next.

Then we compare the numbers that matter to customers and to cash:

  • Manufacturing lead time is about four weeks in both plants.
  • Work in progress, measured at month end, is about ten days in both.
  • On-time delivery is about the same.

So what has the first plant gained? Mostly appearance. The work in progress hasn’t gone; it has been moved to where it is less visible.

Notice another contradiction: ten days of work in progress cannot coexist with a four-week lead time. The explanation is the month-end measurement. Work in progress rises to a mid-month high and is then flushed out in the last week, when release of raw material is choked to make the month-end report look good. The inventory target is met on paper; the working capital need hasn’t changed.

The cost focus that increases costs

The first plant had also invested in automation to reduce direct labour. The labour wasn’t actually released; people were moved to other jobs. So the company now carried the cost of the automation and the same headcount. Total costs had gone up, not down.

A simple test I suggest for any automation proposal: write down the names of the people who will leave and the date they will leave. If the proposal can’t answer that, the savings are probably theoretical.

Meanwhile, the plant’s overtime and emergency freight costs were high, and almost all of it was spent in the last ten days of the month. If bad luck caused these costs, it is remarkably consistent bad luck, because it always strikes at month end.

The hidden waste: making the wrong thing

When we look closer, the biggest waste in both plants isn’t visible on any Lean audit. It is this:

Machines busy making items that aren’t needed yet, while urgent items wait in the queue.

Why does this happen? Because every work centre can see the whole order book and is free to choose what to run next. Naturally, each supervisor picks jobs that keep his machine efficient (long runs, fewer changeovers), without regard for when the item is actually needed.

You can spot the effects easily:

  • Parts piled in front of assembly, yet assembly waiting for one missing part.
  • Some orders finished weeks early and sitting in finished goods, while others are late.
  • Lead times that swing widely from order to order.

In such a plant, local improvements rarely reach the bottom line. Add capacity to a machine through set-up reduction, and it will simply make more of the wrong things sooner. Fix a quality problem near the end of the line, and you may have fixed a problem that was only urgent because time had been wasted earlier.

Get the flow right first

The order of improvement matters:

  1. Fix the flow first. Stop the plant from producing too early or too much. This needs a mechanism: Henry Ford limited space between stations, Toyota limited containers with kanban, and Goldratt’s Drum-Buffer-Rope limits time by releasing material only a fixed time before the due date. Pick the one that fits your environment.
  2. Watch lead time fall and stabilise. With less work in the system, orders move faster and more predictably. Many “capacity problems” disappear, because capacity was being wasted, not lacking.
  3. Now let the system tell you what to improve. Record why orders still run late. A small number of causes will account for most of the delays.
  4. Apply improvement tools to those causes. This is where SMED, Six Sigma, 5S and maintenance programmes earn their keep: on the problems that actually affect flow and delivery.

This approach also answers the question of what not to improve. A quality problem that never delays an order, or a set-up time on a machine with plenty of spare capacity, can wait. Management attention is limited; spend it where it changes results.

Here is what step 3 looks like in practice. Picture a plant that has controlled its release for two months. Each time an order turns red, the supervisor writes one line: which order, where it was stuck, and why. At the end of the month, 60 orders turned red. Sorted by reason:

Reason the order turned red Times
Waiting for painted parts from the outside vendor 24
Machine breakdown at the CNC cell 14
Drawing revision pending from design 9
Rework after final inspection 7
Other (absenteeism, material shortage, etc.) 6

Two reasons account for nearly two-thirds of the delays. The first improvement projects choose themselves: a reliable arrangement with the paint vendor, and a maintenance programme for the CNC cell. The 5S drive in the stores, which never delayed an order, can wait. Next month, the list is made again, and the next project is chosen the same way.

A note on listening to the shop floor

Ask people in a plant with poor flow what their problems are, and you will hear a long list: absenteeism, supplier delays, breakdowns, rejections. All are real. But they are mostly symptoms. Hardly anyone will say, “We waste time and capacity making things too early.” If improvement projects are chosen from the list of symptoms, the company ends up with too many projects and too little effect.

So before launching the next initiative, ask: is our flow under control? If it isn’t, that is the first improvement project, and it may make half the others unnecessary.

Related reading

Mohit Saini

Strategy, supply chain & sales transformation consultant. About · Book a discussion