The Pareto Principle Is More Powerful Than You Think

I was on the floor when an operator ran by me with a bucket. I followed him to the lithographic printing press where I watched him dump the bucket of chemicals into a chiller by hand. I asked why. He said he was having trouble setting his color. This was how they dealt with it.

The top downtime in the factory was color setting time, the top quality defect was color variation, and the top waste was color setting waste. We had the Pareto charts.

Everyone in ops knows the 80/20 rule. Fewer people ask why it shows up everywhere it does.

Here's the core observation: problems create problems.

Pareto's original finding wasn't just a quirky ratio. 80% of the land in Italy, held by 20% of the people.

That's the signature of a specific kind of distribution, one where having more gives you the mechanism to get more.

Wealth begets income. Income begets wealth. That's compounding. Compounding is one of the most reliable ways to generate a Pareto-shaped distribution.

The Pareto distribution applies to your problems because problems cause problems, and solving one problem often solves multiple problems, if you solve the correct problem.

So here's the idea I want to plant:

When you make a Pareto graph, you're not looking for the biggest bar on a chart unless there is a clear Pareto distribution.

When you find a Pareto distribution in your data, you've found the location of something very important to solve.

Suppose I had 100 hours of downtime last month, or a quarter of quality defects, or a stretch of safety near-misses.

Map that loss data across as many dimensions as you can: Man, Machine, Method, Material. Each one gets its own Pareto chart.

Most dimensions will show some skew. There is always a biggest bar. That's normal, real data is messy.

What you're hunting for is the dimension where the skew is sharpest relative to the others. The sharper the skew, the higher the leverage of digging there.

Finding the skew is only step one.

Step two is going to the floor and learning what problems are happening in that process/location.

Sometimes it's one clear issue doing steady damage. It could be a machine issue or a SKU that doesn't have a good setup.

Sometimes there's just variation causing a cascade of other problems that seem almost unrelated.

If you take all your loss data (safety, quality, production, etc), you may find that doing multi-dimensional Pareto analysis on each directs you to a truth many leaders never figure out - your issues are all coming from common epicenters.

Back to my story, I built a controller to hold press chemical concentration precisely and automatically.

It solved the top quality, waste, and downtime issues all at the same time.

Three separate measurement systems. Three separate top bars. One cause: simple variation on a dosing pump, previously invisible, driving all three.

The charts tell you where to look. You still have to go look.

What have you seen that made you ask, "Why are we doing that?"