You’ve seen this play out. A plant is underperforming. Someone brings in outside help. There’s a flurry of activity — assessments, Kaizen events, new dashboards, a big binder of recommendations. OEE ticks up for a few weeks. Then the consultant leaves.
Three months later, the plant is back where it started. Same problems. Same Pareto chart. Same morning meeting. The binder is in a drawer.
This pattern is so common it’s almost expected. But it’s not inevitable. It happens for specific, diagnosable reasons.
Recommendations without routines
A recommendation is not an intervention. “Reduce changeover time on Line 3” is a recommendation. A routine looks like this:
- SMED analysis completed. Internal/external activities classified.
- Standard work document created with time targets per step.
- One Point Lesson trained across all crews.
- Weekly changeover time audit built into shift supervisor’s review.
- 30/60/90-day recurrence check scheduled with named owner.
The recommendation tells you what to do. The routine tells you who does it, when they do it, how they know if it’s working, and what happens when it stops working.
Most turnarounds ship recommendations. The plants that hold ship routines.
Fixes without proof they held
Here’s the pattern: a CI team identifies a problem. They implement a fix. OEE improves for two weeks. Everyone declares victory and moves to the next project.
Nobody checks at 30 days. Nobody checks at 60 days. At 90 days, the problem is back — but now it’s “a new problem” because nobody connected it to the fix that didn’t hold.
Recurrence tracking is the most boring, least glamorous part of continuous improvement. It’s also the part that determines whether your improvement program compounds or just cycles.
Without it, you’re not doing continuous improvement. You’re doing continuous activity.
Diagnosis without decomposition
The typical turnaround starts with a Pareto chart. “Labeler downtime is our top loss.” Everyone nods. CI resources point at the labeler.
But labeler downtime caused by an untrained operator on second shift is a completely different problem than labeler downtime caused by a worn cam follower. Same bar on the Pareto. Different root cause. Different fix. Different owner.
Without decomposing where the variance lives — within-shift vs. between-crew vs. schedule-induced vs. equipment — the turnaround team chases the symptom. They fix the visible thing. The underlying driver reasserts itself in three weeks. The Pareto chart doesn’t change. Morale erodes.
The turnaround didn’t fail because the team was bad. It failed because the diagnosis was incomplete.
No economic translation
Most turnaround plans rank interventions by severity: worst OEE first, biggest Pareto bar first, loudest complaint first.
But severity and return are different things. A line at 55% OEE with $80K in annual exposure and a $5K fix is a better investment than a line at 48% OEE with $15K in exposure and a $20K fix. The first one pays back 16:1. The second one doesn’t even break even in year one.
When interventions aren’t ranked by economic return, CI resources go where the noise is loudest, not where the money is. The CFO can’t connect the improvement program to the P&L. Funding gets cut. The turnaround dies of budget, not capability.
What holds
The turnarounds that stick share four traits:
- The truth is stabilized first. Before any intervention, validate rates, planned downtime rules, reason codes, and data quality. You can’t diagnose a plant you can’t measure.
- Variance is decomposed, not averaged. Every line gets classified by reliability, direction, and economic exposure — not just by OEE rank.
- The right constraint gets attacked. TPM, SMED, standard work, schedule changes — chosen based on where the instability pattern actually lives, not based on which methodology someone is certified in.
- The fix is proved. Recurrence tracking at 30/60/90 days. Before/after documented. If the fix didn’t hold, you know it before the P&L does.
A plant does not deploy formulas. It deploys routines. The turnaround that builds the routine — complete with ownership, measurement, and recurrence verification — is the one that holds after the consultant leaves.
The 10-Day Throughput Reliability Assessment is designed as the starting point for turnarounds that hold. Not recommendations — decomposed diagnosis, economic ranking, and a structured intervention plan with named owners and recurrence tracking built in.