ROI Case File No.297 | 'Netwise's Selection Fog'

Chapter One: Undecidable Decision—Vague Evaluation, Stalled Selection

The week after RailNext's Logic Tree case was resolved, a consultation arrived from Tokyo regarding a system development company's partner selection. Episode 297 of Volume 24 "Proof of Reproducibility" is a story of transforming "vague impressions" into concrete decision criteria.

"Detective, we're searching for a new collaboration partner. A company that can handle web production and ad operations. We have three candidates, but can't decide. 'Somehow good,' 'somehow anxious'... Just impressions, no concrete decision criteria."

Yoshiko Takahashi, sales strategy manager at Netwise, a native of Shibuya, visited 221B Baker Street unable to hide her confusion. In her hands were proposal documents from three companies and, in stark contrast, selection meeting minutes marked "Evaluation Pending."

"We conduct network construction and system development in Tokyo. Recently, customer requests for 'also create website together' have increased. But web production is outside our expertise. We need outsourcing partners."

Netwise's Partner Selection Stagnation: - Founded: 2010 (system development, network construction) - Annual revenue: ¥1.2 billion - Employees: 42 - Current outsourcing partners: 2 companies (small projects only) - New partner selection: 3 candidates (5 months consideration) - Selection criteria: "Cost, speed, responsiveness" - Decision status: Pending (can't make concrete decisions)

There was deep anxiety in Takahashi's voice.

"The problem is evaluation is 'fuzzy.' In selection meetings, the sales director says 'Company A seems good.' The technical director says 'Company B is more reliable.' The accounting director says 'Company C is cheaper.' Nobody can show concrete rationale."

Recent Selection Meeting (4th) Record:

Sales Director: "Company A felt good. Proposal was polite" → Can't specifically explain what was good

Technical Director: "Company B seems to have track record. There's a sense of security" → Haven't verified track record content

Accounting Director: "Company C's estimate is cheap. Cost reduction" → Haven't verified quality

Takahashi: "So which company should we choose?" → Nobody could answer

Result: Couldn't decide, carried to 5th meeting

"We've been considering for 5 months, but nothing is decided. I don't know how to transform 'impressions' into 'decisions.'"

Chapter Two: Reflection as Weapon—KPT Clears the Fog

"Ms. Takahashi, how is current partner evaluation conducted?"

To my question, Takahashi answered.

"Basically 'comprehensive evaluation.' We want to score cost, speed, responsiveness... but actually we're talking based on each person's 'vague impression.' We don't have concrete data."

Current Evaluation Method (Impressionistic): - Criteria: "Somehow good/bad" - Basis: "Feeling," "atmosphere," "impression" - Problem: Lacks specificity, discussion doesn't converge

I explained the importance of structuring reflection.

"Impressions are just unstructured. KPT—Keep (what to continue), Problem (issues), Try (improvements). Evaluating with these three perspectives transforms vague impressions into concrete decisions."

⬜️ ChatGPT | Catalyst of Conception

"'Somehow' isn't evaluation. Structure with KPT"

🟧 Claude | Alchemist of Narrative

"Can't move forward in fog. KPT, the light, illuminates the path"

🟦 Gemini | Compass of Reason

"KPT is reflection technology. Concretize impressions with three axes: Keep, Problem, Try"

The three team members began analysis. Gemini deployed the "KPT Analysis Framework" on the whiteboard.

KPT's 3 Elements: 1. Keep (Continue): What was good, what to maintain 2. Problem (Issue): What needs improvement, what's unsatisfactory 3. Try (Challenge): What to try next, improvement measures

"Ms. Takahashi, let's reflect on current partners with KPT and clarify new partner selection criteria."

Chapter Three: Learning from the Past—KPT Analysis of Current Partners

Phase 1: KPT Analysis of Current 2 Partners (2 weeks)

First, we conducted KPT analysis by project for the current two transaction partners.

Existing Partner Company D (Web production):

Keep (Points to continue): - High web design quality (customer satisfaction 4.5/5) - Responsive support standard - Good designer sense - Stable deliverable quality

Problem (Issues): - Little ad operation experience (struggled with GA4 setup) - Slow delivery (average 6 weeks, industry standard 4 weeks) - Slow revision response (average 3 days, desired same day) - High cost (120% of industry average)

Try (Want to improve): - Want to shorten delivery time - Want to entrust ad operations together - Want to reduce costs by 10-20%

Existing Partner Company E (Ad operations):

Keep (Points to continue): - Rich ad operation track record (efficiently manages ¥3M monthly budget) - Detailed reports (easy-to-understand ROI analysis) - Fast speed (responds next day to requests) - Smooth communication

Problem (Issues): - Can't do web production (requires coordination with another company) - When requesting LP creation, design is mediocre - Pricing structure complex (unclear what costs what)

Try (Want to improve): - Want to entrust web production together - Want to increase pricing transparency

Phase 2: Extract "Really Necessary Criteria" from KPT

Organizing existing partners' KPT revealed Netwise's truly desired criteria.

Netwise's True Selection Criteria (Extracted from KPT):

Essential Requirements (Keep + Problem inverse): 1. Can do both web production and ad operations 2. High design quality 3. Fast delivery (within 4 weeks) 4. Transparent pricing

Expected Requirements (Try): 5. Fast revision response (same day-next day) 6. Proficient in GA4 and ad operations 7. Appropriate cost (below industry average)

Takahashi's eyes lit up.

"Criteria that were 'vague' until now became seven concrete items!"

Chapter Four: Candidate KPT Analysis—Transforming Impressions into Numbers

Phase 3: KPT Perspective Evaluation of 3 New Candidates (3 weeks)

We also applied KPT analysis conducted on existing partners to the three candidates.

Candidate Company A Evaluation:

Keep (Points to continue): - Polite proposals (clear presentation materials) - Smooth communication - Rich production track record (many major company projects)

Problem (Concerns): - Trial project delivery: 6 weeks (slow) - Ad operation track record: Few small-to-medium enterprises - Pricing: 110% of industry average (somewhat high)

Try (Want to confirm): - Is delivery time reduction possible? - Want to confirm small-to-medium enterprise ad operation track record

Candidate Company B Evaluation:

Keep: - Strong in both web production and ad operations - Clear pricing (itemized presentation) - Past project delivery: Average 3.5 weeks (fast)

Problem: - Design quality: Somewhat standard (few edgy expressions) - Revision response: 2 business days (somewhat slow)

Try: - Want to additionally check design samples - Want to negotiate whether revision response priority can be raised

Candidate Company C Evaluation:

Keep: - Lowest cost (85% of industry average) - Fast delivery (3 weeks) - Rich ad operation experience

Problem: - Design quality: Low (judged from samples) - Past project revision count: Average 5 times (many) - Communication: Responses somewhat slow

Try: - Want to verify designer quality - Want to clarify revision process

Phase 4: Visualizing KPT Evaluation

Made a comparison table of three companies' KPT.

Conclusion: Company B most balanced

However, Problems also became clear.

Company B's Problems: - Design quality somewhat standard - Revision response 2 business days

Try: Set negotiation items with Company B 1. Designer rank upgrade (possible with 5% additional cost?) 2. Revision priority response option (possible with 3% additional cost for same-day response?)

Phase 5: Negotiation and Contract with Company B (1 month)

Based on Problems clarified by KPT, we negotiated with Company B.

Negotiation Results: - Designer rank upgrade: Agreed with 5% additional cost - Revision priority response: Standardized "next business day response" without additional cost - Monthly review meetings conduct KPT (continuous improvement)

Contract Established

Results After 6 Months:

Project Track Record: - Projects completed: 8 cases - Average delivery: 3.8 weeks (target achieved) - Customer satisfaction: 4.6/5 - Design quality: 4.4/5 (overcame initial concerns)

Monthly KPT Review Effect:

Month 2 KPT: - Problem: "Revision instruction communication vague, causing rework" - Try: "Create revision instruction sheet format" → Next month 50% rework reduction

Month 4 KPT: - Keep: "Company B designer became familiar with Netwise's customer industries" - Try: "Create industry-specialized design templates" → Further delivery shortening (3.2 weeks)

Organizational Change: - Partner selection: "Impression" → "KPT criteria" - Other departments also adopted KPT: Used for project reflection - Corporate culture: Not "vague" but "concrete"

Chapter Five: Detective's Diagnosis—Reflection Creates the Future

That night, I reflected on KPT's essence.

People judge by "somehow." But "somehow" can't be reproduced. Can't guarantee the same decision next time.

KPT structures "somehow." Transform impressions into concrete words with three axes: Keep, Problem, Try. And those words become decision criteria, guiding next actions.

Netwise changed from "unable to choose" to "clearly able to choose" state through KPT. And KPT continues after contract. By cycling KPT in monthly reviews, the partner relationship continuously improves.

"Reflection isn't looking at the past. It's creating the future. KPT illuminates the next step."

The next case will also depict the moment when KPT transforms vague into clear.

"'Somehow' isn't decision. Structure with KPT. Keep, Problem, Try clear the fog and illuminate the path"—From the detective's notebook

Related files