The Pareto chart is the most trusted tool in continuous improvement. It’s also the one most likely to send you to the wrong fix.
Not because it’s wrong. It does exactly what it’s designed to do: rank losses by frequency or duration. “Labeler downtime” is the tallest bar. Send CI to the labeler.
Here’s the problem: the Pareto doesn’t tell you why the labeler keeps going down. And the why determines everything.
Same bar, different problem
Labeler downtime caused by an untrained operator is a completely different problem than labeler downtime caused by a worn cam follower.
- Untrained operator: The operator isn’t following centerline settings. They trigger faults that experienced operators avoid. The fix is training and standard work. Production owns it.
- Worn cam follower: The equipment is degrading. The labeler faults because a component is past its service life. The fix is a PM interval adjustment and parts replacement. Maintenance owns it.
- Brutal product mix: The labeler faults because it just ran 14 changeovers in five days. Short runs never let it stabilize. The fix is schedule optimization and product sequencing. Planning owns it.
Three different root causes. Three different fixes. Three different owners. Same bar on the Pareto chart.
If you don’t decompose which one it is, you’ll default to the most visible explanation (usually maintenance), implement the wrong fix, and three weeks later the Pareto chart looks exactly the same.
The variance tells you more than the total
The Pareto ranks losses by how much. What you actually need to know is where the variance lives:
Within-shift variance — the line degrades during a single shift. Faults, jams, material quality issues. This is an equipment or material problem. Maintenance owns it.
Between-crew variance — first shift runs 80%. Second shift runs 62%. Same equipment, same product, same day. This is a training or standard work problem. Production owns it.
Schedule-induced variance — the line runs great on long runs and falls apart on short runs. This is a changeover or product mix problem. Planning owns it.
Once you know where the variance lives, the fix is obvious. The Pareto can’t get you there because it aggregates everything into one bar.
Severity vs. return
There’s a second way the Pareto misleads: it ranks by severity, not by economic return.
Consider two lines, both with bad OEE:
- Line A: 55% OEE, $80K annual exposure, fix costs $5K. Return: 16:1.
- Line B: 48% OEE, $15K annual exposure, fix costs $20K. Return: 0.75:1.
The Pareto sends you to Line B because 48% is worse than 55%. But Line A has 5x the exposure and the fix costs a quarter as much. Line A is the higher-return investment by a factor of 21x.
This isn’t theoretical. In a recent four-line assessment, the highest-return intervention ($280K annual recovery at $15–25K cost) would have been missed entirely by OEE-based prioritization. The traditional Pareto would have sent CI to a different line.
What to do instead
The Pareto chart is a starting point, not a decision tool. Use it to identify which losses to investigate. Then decompose:
- Where does the variance live? Within-shift, between-crew, or schedule-induced? Each one points to a different owner.
- What does the instability cost? Not just OEE points — dollars. Overtime, expediting, safety stock, quality events, lost throughput.
- What’s the return on fixing it? Rank interventions by economic return, not severity. The $5K fix with $80K recovery beats the $20K fix with $15K recovery every time.
The Pareto tells you where to look. Variance decomposition tells you what to do. Economic translation tells you whether it’s worth doing.
That’s the sequence. Skip any step and you end up back at the same morning meeting with the same chart.
The 10-Day Throughput Reliability Assessment includes variance decomposition, economic exposure by line, and interventions ranked by return. Start with one line →