Three-Layer Decomposition — The Three Wastes at the Bottleneck

System properties set the statistics of flow — variability accumulates through dependency, so WIP inevitably piles up in front of the true constraint

Station A
input / prep
→
Station B
BOTTLENECK — sets system output
→
Station C
sub-processing
→
Station D
assembly
→
Station E
final inspection

⬆ WIP piles up in front of B

1 hour lost at B = 1 hour lost system-wide

Layer 2 Conflict Level — "How do we fully exploit B?" — and hit a mental limitation: not knowing how

The three wastes — where TOC parts ways with convention; each TOC column IS the subordination demanded by Step 3 of the Focusing Steps

Waste Conventional Thinking TOC Approach (The Goal) Layer-3 Choke Point: Subordination
① Processing defective parts
B's scarce time is spent on scrap		 · Reduce defect rate
· Pareto / fishbone charts
· The broader the scope, the better (whole line)		 · No wasted work at B
· 100% inspection before B
· Focus on the bottleneck ONLY (with flow through B kept smooth, everyday yield improves faster)		 · Inspectors drop line-wide quality standards; serve B's spec
· Upstream swaps its own yield KPI for kit-completeness before B
Inspectors drop line-wide quality standards; serve B's spec. Upstream swaps its own yield KPI for kit-completeness before B.
② Waiting for material / idle time · Stock buffers / safety stock
· Prep early; forecast long leads
· Each station chases its dates		 · Mealtime relief shifts (B runs on)
· Full-kit release to B
· One concentrated buffer (DBR) replaces scattered stock		 · Upstream stops racing ahead — release follows B's drumbeat
· Lead drops "shortages = purchasing"
Upstream stops racing ahead — release follows B's drumbeat. Lead drops "shortages = purchasing".
③ Making what isn't needed · Absorb frequent customer changes
· Build ahead / forecast schedules
· Spread multi-skilled prep wide		 · Make only what is needed NOW
· One single priority queue before B
· Non-bottlenecks STOP speculative future orders		 · Drop the "an order might come" assumption — stop now
· Planners stop build-ahead cover
Drop the "an order might come" assumption — stop now. Planners stop build-ahead cover.
Layer 3 Assumption Level — the core problem: the Local-Efficiency Fallacy

Every Layer-2 subordination asks non-bottlenecks to give up local efficiency — and the Layer-3 assumption forbids exactly that

Where the limitation bites:

Non-bottlenecks give up local efficiency ↓ Conventional KPIs refuse — every station must stay busy ↓ Step 3 "Subordinate" runs in name only — F5 collapses into a one-point fix

Countermeasure — don't run F5 on a limitation; let it be SEEN:

  • Inventory piled up in front of B
  • "Efficient" non-bottlenecks making output nobody needs
  • The hidden cost of firefighting heroes

Loop: limitation unloosened → Step 3 runs in name only → F5 degenerates. Limitation seen → subordination actually happens → F5 is the holistic solution.

Beneath Layer 3 sits "who is willing to see" — trust position. Change immunity can freeze the logic right there.

How to use this chart