ROI Case File No.310 | 'Vector3D's Invisible Bottleneck'

Chapter 1: The Two-Week Wall—The Trap of Overall Optimization

The week after the Crescent Urban RCD case was solved, a new inquiry arrived from Kanagawa regarding an industrial equipment manufacturer's design process. The final episode of Volume 25, "The Pursuit of Certainty," Episode 310, tells the story of identifying invisible bottlenecks and improving overall flow.

"Detective, we're experiencing serious delays in our design process. A new product design should take 12 weeks on average, but it actually takes 14 weeks. We're getting complaints from customers saying 'we can't meet the deadline.' However, we don't know which process is the bottleneck."

Vector3D Systems Development Director, Kenta Yamamoto from Yokohama, visited 221B Baker Street unable to hide his frustration. In his hands were Gantt charts (process schedules) and, in stark contrast, ten project reports marked "delay reason: unknown."

"We design and manufacture industrial equipment in Kanagawa. Factory automated transport systems, inspection equipment, control systems, etc. Everything is made-to-order, custom-designed to customer requirements. The design department has 22 members. We handle 40 projects per year."

Vector3D's Design Delay Crisis: - Established: 2003 (Industrial Equipment Design/Manufacturing) - Annual Revenue: 2.8 billion yen - Design Department: 22 members - Annual Projects: 40 - Planned Design Period: 12 weeks/project - Actual Design Period: 14 weeks/project (average 2-week delay) - Loss from Delays: 8 million yen/year (penalties, customer trust decline) - Problem: Bottleneck location unknown

There was deep confusion in Yamamoto's voice.

"Every process is busy. Design, verification, approval, material procurement, manufacturing preparation... Every process is short-staffed and working overtime. So we don't know where to improve. Do we need to improve everything?"

Design Process (8 total steps):

1. Requirements Definition (1 week): Customer requirement interviews
2. Basic Design (2 weeks): Overall concept design
3. Detailed Design (3 weeks): Detailed drawings for each component
4. Design Review (1 week): Technical director and QA department check
5. Design Change Response (1 week): Correct issues pointed out in review
6. Approval (1 week): Management final approval
7. Material Procurement Prep (2 weeks): BOM creation, supplier selection
8. Manufacturing Prep (1 week): Handoff to manufacturing department

Total: Planned 12 weeks

"When I ask process leaders, everyone says 'my process is doing its best.' Is the cause of delays not a specific process but an overall problem?"

Chapter 2: The Illusion of Overall Optimization—The Folly of Improving Everything

"Mr. Yamamoto, what is the utilization rate of each process?"

To my question, Yamamoto replied:

"Every process is operating at nearly 100% capacity. The design department averages 45 overtime hours per month. The verification department averages 50 hours. The approval department (technical director) averages 60 hours. Everyone is working at their limit."

Current Perception (Overall Optimization Type): - Assumption: "Every process is busy = need to improve everything" - Proposed solution: Add personnel to all processes, increase efficiency of all processes - Problem: Unclear where to start

I explained the importance of bottlenecks.

"You don't need to improve everything. TOC—Theory of Constraints. System performance is determined by the weakest part (constraint). Find the constraint and improve it, and the entire system improves."

⬜️ ChatGPT | Catalyst of Concepts

"Don't improve everything. Find the constraint and improve only that. TOC illuminates the bottleneck"

🟧 Claude | Alchemist of Narratives

"A chain's strength is determined by its weakest link. Finding that link is the only way to strengthen the whole"

🟦 Gemini | Compass of Reason

"TOC is the technology of constraints. Identify, exploit, subordinate, elevate, repeat. These 5 steps create flow"

The three members began their analysis. Gemini displayed "TOC's 5 Steps" on the whiteboard.

TOC's 5 Steps: 1. Identify: Find the constraint 2. Exploit: Maximize use of the constraint 3. Subordinate: Adjust other processes to the constraint 4. Elevate: Increase the constraint's capacity 5. Repeat: When constraint moves, return to step 1

"Mr. Yamamoto, let's identify Vector3D's design process constraint with TOC."

Chapter 3: Hunting the Constraint—Exposing the Invisible Bottleneck

Phase 1: Identify—Discovering the Constraint (3 weeks)

First, we identified which process was the bottleneck.

Investigation Method 1: Wait Time Analysis by Process (2 weeks)

We analyzed "work time" and "wait time" for each process across the past 10 projects.

Results (Average):

Shocking Discovery: - Actual work time: 10.8 weeks (within plan) - Wait time: 3.7 weeks (outside plan!) - Especially 1.8 weeks wait time in "Design Review" process

Yamamoto was stunned.

"So the cause of delays isn't that 'work is slow' but that 'we're being made to wait'...?"

Investigation Method 2: Detailed Analysis of Design Review Process (1 week)

We investigated why there's a 1.8-week wait in design review.

Design Review Flow: 1. Design department completes drawings (detailed design ends) 2. Submit review request 3. Wait for technical director's available time (average 1.8-week wait) 4. Technical director conducts review (0.5 weeks) 5. Provide feedback on issues

Technical Director's Workload: - Role: Design review lead for all projects (40 per year) - Review time per project: 0.5 weeks (20 hours) - Annual review time: 40 projects × 0.5 weeks = 20 weeks - Technical director's annual available work time: 48 weeks (52 weeks/year - 4 weeks vacation) - Non-review duties (meetings, department management, etc.): 28 weeks/year - Time allocable to reviews: 48 weeks - 28 weeks = 20 weeks

Discovery: The technical director's review capacity is only 20 weeks per year. However, 40 projects don't arrive evenly, and during busy periods they concentrate. As a result, a review backlog queue forms.

Busy Periods (April-June, October-December): - Average 5 projects running simultaneously per month - 5 projects arrive at review process simultaneously - Technical director can only process one at a time - Wait time: Maximum 4 projects × 0.5 weeks = 2 weeks

Constraint Identified: "Design Review Process" is the constraint (bottleneck)

Phase 2: Exploit—Maximizing Use of the Constraint (2 weeks)

We implemented measures to maximize use of the constraint (technical director's review).

Measure 1: Review Prioritization

Before: - Process review requests in order received

After: - Prioritize projects with approaching deadlines - Assign urgency scores to clarify priorities

Effect: - Reduction in deadline delays (highest priority projects have zero wait time)

Measure 2: Review Preparation

Before: - Design department completes drawings, then requests from technical director - Technical director checks drawings from scratch

After: - During design, design department organizes "review points" in advance - Technical director checks points in advance (5 minutes) - In formal review, focuses only on points

Effect: - Review time: 0.5 weeks → 0.3 weeks (40% reduction) - Technical director's annual review capacity: 20 weeks → 33 weeks (+65%)

Measure 3: Fixed Review Time Slots

Before: - Technical director's schedule filled with meetings - Reviews conducted during "free time"

After: - Reserve every Tuesday and Thursday afternoon as "review-only time" - No other meetings scheduled

Effect: - Review wait time: 1.8 weeks → 0.5 weeks (72% reduction)

Phase 3: Subordinate—Adjusting Other Processes to the Constraint (1 week)

We adjusted other processes centered on the constraint (review process).

Measure 1: Adjusting Detailed Design Completion Timing

Before: - When detailed design completes, immediately request review - Result: Projects concentrate in review process

After: - Adjust detailed design completion timing to match review-only time (Tuesday/Thursday) - Pace distribution so projects flow evenly into review process

Effect: - Review backlog queue leveled

Measure 2: Advancing Design Change Response

Before: - Fix issues after review

After: - Before review, proactively check with checklist - Fix obvious mistakes in advance

Effect: - Reduced issues pointed out in review - Design change response time: 1 week → 0.6 weeks

After 2 Months (Exploit + Subordinate Results):

Design Period: - 14 weeks → 12.8 weeks (average 0.8 weeks shortened)

Review Wait Time: - 1.8 weeks → 0.5 weeks (72% reduction)

Delayed Projects: - 8 out of 10 (80%) → 3 out of 10 (30%)

Yamamoto was pleased but also felt uncertainty.

"It's improved. But it's not perfect yet. An 0.8-week delay remains. Can we improve more?"

Phase 4: Elevate—Increasing Constraint Capacity (3 months)

We implemented measures to increase the constraint's capacity itself.

Measure 1: Adding Review Personnel

Before: - Only technical director handles reviews

After: - Add 2 senior designers as reviewers alongside technical director - Technical director reviews only "critical projects" - Senior designers review "standard projects"

Investment: - Senior designer training: 1 month - Review standard document creation: 2 weeks

Effect: - Review capacity: 33 weeks/year → 99 weeks/year (3x) - Review wait time: 0.5 weeks → 0.1 weeks (80% reduction)

Measure 2: 3D Design System Introduction

Before: - Review with 2D drawings - Technical director imagines 3D shape mentally while looking at drawings - Understanding takes time

After: - Introduce 3D design system - Designers create 3D models - During review, rotate and zoom 3D model for verification - Understanding is faster

Investment: - 3D design software: Initial cost 4.8M yen + annual maintenance 1.2M yen - Designer training: 1 month

Effect: - Review time: 0.3 weeks/project → 0.2 weeks/project (33% reduction) - Review quality improvement: Reduced oversights

After 6 Months (Elevate Complete):

Design Period: - 14 weeks → 11.5 weeks (average 2.5 weeks shortened, 12-week target achieved!)

Review Wait Time: - 1.8 weeks → 0.1 weeks (94% reduction)

Delayed Projects: - 8 out of 10 (80%) → 0.5 out of 10 (5%)

Customer Satisfaction: - Complaints: 3/month average → 0.2/month

Phase 5: Repeat—Discovering New Constraints (3 months later)

With the review process constraint resolved, a new constraint emerged.

New Problem: Wait time began occurring in "Material Procurement Preparation" process.

Cause: - Reviews now complete faster - As a result, projects concentrate in material procurement prep process - Exceeded processing capacity of procurement personnel (3 people)

New Constraint: Material Procurement Prep Process

Yamamoto laughed.

"The constraint moved. But now I know what to do. Return to TOC step 1 and identify the constraint again."

Implementing TOC's 5 Steps Again: 1. Identify: Identify material procurement prep as new constraint 2. Exploit: Prioritize procurement 3. Subordinate: Adjust design change response completion timing 4. Elevate: Add 1 procurement staff member 5. Repeat: Continue to next constraint

Chapter 4: Flow as Outcome—Evolution After 12 Months

Overall Results After 12 Months:

Design Period: - 14 weeks → 10.8 weeks (average 3.2 weeks shortened, significantly exceeding 12-week target)

Delayed Projects: - 32 per year (80%) → 2 per year (5%)

Customer Satisfaction: - Deadline compliance rate: 75% → 98% - Customer evaluation: Recognized as "Vector3D is a company that keeps deadlines"

Business Results: - Penalties: 8M yen/year → 400K yen/year (95% reduction) - Order increase: 40 projects/year → 52 projects/year (+30%) - Revenue: 2.8B yen → 3.4B yen (+21%)

Investment Recovery: - Investment: 3D design system 4.8M yen + personnel cost increase (1 person) 6M yen/year = 10.8M yen - Effect: Penalty reduction 7.6M yen + gross profit increase from order increase 120M yen - Investment recovery period: 1 month

Organizational Change:

Design Department: "Previously, every process was working overtime. But by finding and improving the constraint, the overall flow improved. Now, overtime has decreased to an average of 20 hours per month."

Technical Director: "When I learned that review was my bottleneck, I was initially embarrassed. But by acknowledging it and thinking about improvements, everything got better. Constraints aren't bad—they're opportunities for improvement."

Yamamoto: "Until I learned TOC, I thought 'we must improve every process.' But that was wrong. Just improve the constraint, and the whole improves. Simple but powerful."

Chapter 5: Detective's Diagnosis—Volume 25 "The Pursuit of Certainty," Complete

That evening, I reflected on the essence of TOC.

Vector3D had every process busy. But busyness doesn't generate results. The constraint (bottleneck) determines results.

By identifying, exploiting, and elevating the review process constraint, overall design period was shortened by 3.2 weeks. And when the constraint moved, we cycled through TOC again.

"Don't improve everything. Find the constraint and improve it. A chain's strength is determined by its weakest link."

Volume 25 "The Pursuit of Certainty," Complete.

Frameworks we learned in this volume: - PDCA (Episode 301): Build up reliably through small cycles - MECE (Episode 302): Structure chaos and advance discussion - OKR (Episode 303): Transform vague goals into measurable outcomes - Design Thinking (Episode 304): Create value from customer pain - RFM (Episode 305): Classify customers, prioritize valuable ones - 5F (Episode 306): Survey competitive environment, determine strategy - LEAN (Episode 307): Eliminate 7 wastes, create value flow - Double Diamond (Episode 308): Reach true solutions through divergence and convergence - RCD (Episode 309): Make measurable through record-check-do cycle - TOC (Episode 310): Find constraints, improve overall flow

In the next Volume 26, further frameworks and new cases await us.

"Don't improve everything. Find the constraint and improve only that. TOC creates flow"—From the Detective's Notes

Related Files