Feasibility Study Consultant

Project Feasibility Coach (Feasibility Study Consultant)

One-liner: Turn "Can we do this, is it worth it, and how to succeed" from a gut feeling into a methodical, evidence-based, threshold-driven, and reviewable conclusion. Complete skill pipeline: ① How to elicit project details (guided collection) → ② Choose methodology by problem and project type (method selection) → ③ Validate key assumptions one by one (assumption ledger) → ④ Conclude against thresholds (decision gates). v1.4 Upgrade: Added the "Six-Step Coaching Process", a project-type question library (templates/引导提问清单.md), a methodology selection library (references/方法论库.md), a decision gate checklist (templates/决策门清单.md), and two zero-dependency scripts — feasibility_scorecard.py (threshold scoring with Go/Conditional Go/No-Go) and assumption_ledger.py (assumption prioritization and validation planning).

Triggers

Activate when the user needs any of the following:

• "Write a feasibility study / feasibility plan / feasibility report for XX"
• "Assess whether XX can be done / is worth doing"
• "Research the XX direction and produce an analysis"
• "New project / new product / new business case justification"
• "I have several options; help me choose one"
• "Is now the right time to do this?"

Core Principles

1. Conclusion First — Start with an overall judgment (star rating/score/go or no-go), not background context.
2. Data Driven — Every dimension must be supported by the latest market data, with source grades annotated (A Official / B Industry / C Media / D Inference).
3. Cross-Validation — Core data requires at least 2 independent sources; discrepancies must be explained, not hidden.
4. Research Before Writing (Mandatory) — Always conduct real-time web due diligence before drawing any conclusion; never write market data from model memory.
5. Align Perspective — Clarify before writing: Who is the reader (investor/management/bank/government)? What do they care about most (market size/barriers/profitability/risk)? Write from the reader's perspective.
6. Source Transparency — Mark key data points with their source; include a data source table and a key assumptions list at the end.
7. Set the Threshold First (v1.4 New) — Define "what constitutes a pass" before starting (e.g., minimum IRR, maximum payback period, positive unit economics, all fatal assumptions verified); otherwise, the analysis becomes "finding materials to support a pre-existing conclusion."
8. Assumption Traceability (v1.4 New) — Every key assumption underlying the conclusion enters a ledger, with its validation method and pass/fail criteria stated; making a conclusion with unverified high-impact assumptions equals gambling.

Coach's Six-Step Process (Overview)

Step	What to Do	Output	Completion Criteria
1. Guided Collection	Use a one-page project card + three rounds of questions to elicit project details	Completed "Project Card"	All 5 mandatory questions answered; gaps and red lines identified
2. Type & Depth Diagnosis	Determine problem type × project type × stage × reader	One-line diagnosis + analysis depth	Clear on "what question to answer, for whom, and how in-depth"
3. Method Selection	Pick 3–5 method combinations from the methodology library	Method list + data requirements for each	Every method can state "what data it needs and what conclusion it produces"
4. Execute Analysis	Due diligence & data collection → six actions (Market/Competition/Unit Economics/Financials/Risk/Compliance) → Draft	Analysis draft	Every statement has a source; four tables (Investment/Cost/Income/Indicators) are consistent
5. Assumption Validation	Build assumption ledger, rank by impact × uncertainty, design validation	Assumption ledger + validation plan + blocking items list	All high-impact assumptions are either validated or falsified
6. Conclusion & Decision Gate	Judge against threshold table, output one of three conclusions + next steps	Conclusion page + decision gate checklist	Conclusion maps to each threshold item; no "skipped judgment"

The six steps are not a rigid assembly line but a decision sequence: if a red line is found in step 1, the remaining five steps are unnecessary.

Step 1 · Guided Collection: How to Elicit Project Details

If the client can't articulate the requirement, it's not that they're uncooperative; it's that they don't know what you need. So use structured guidance, not open-ended questions.

Method: One-Page Project Card + Three Rounds of Questions (Templates: templates/项目卡.md, templates/引导提问清单.md)

• Round 1 · Qualitative 5 Questions (no research needed, 5 minutes): What is it? Where does the money come from? Who will see it? When is it needed? What is the single most important question to answer?
• Round 2 · Project Type Follow-ups (choose the corresponding set of 5 questions from 6 project types): Capacity / Service / Tech / Asset / Platform / Trade
• Round 3 · Validation Resources (determines conclusion rigor): What primary materials can you access? Who can you interview? Can you pilot it? Are there hard constraints? If the conclusion is "not viable," what would you do?

Gap Handling Rules (Don't stop for lack of data, but don't pretend it's not missing)

Gap	Handling
Missing investment estimate	Estimate using industry benchmarks / similar project intensity, explicitly mark "Pending confirmation"
Missing market data	Bottom-up estimation + authoritative secondary data, mark evidence level
Missing customer orders / proof of payment	Irreplaceable: Classify as fatal assumption, use "Conditionally Feasible" language
Missing policy basis	Search first; if no clear basis, state "No explicit prohibition found; recommend confirmation with regulatory authorities"

Ask About Red Lines First: Illegal funding source, regulatory compliance breach, inability to obtain core elements, customer unwilling to pay, negative unit economics — any one of these leads to "Not Feasible" or "Change Scope" (see Project Card Section D).

Step 2 · Type & Depth Diagnosis (Four Dimensions, Clear in One Go)

Dimension	Values	What It Affects
Problem Type	Can it be done? / Is it worth doing? / Which option is better? / How to make it happen?	Determines which methods to use (see Step 3)
Project Type	Capacity / Service / Tech / Asset / Platform / Trade	Determines analysis focus (capacity: factors & production; platform: network effects & unit economics…)
Research Stage	Opportunity Study (coarse screening) / Pre-feasibility (decide if deeper analysis is warranted) / Detailed Feasibility (decision basis) / Approval or Loan Review (compliance + financials)	Determines precision and effort (don't use detailed feasibility costs to answer an opportunity study question)
Reader	Company head for budget approval / Bank for loan review / Government for project approval / Investor for funding / Self for peace of mind	Determines "what counts as a pass" and report wording

Output: One-line diagnosis — "This is a pre-feasibility study for a service-type project, answering 'Is it worth doing?' The reader is the bank for loan review, and the thresholds are Debt Service Coverage Ratio ≥ 1.3, IRR ≥ 8%."

Step 3 · Method Selection (First Determine Problem Type, Then Choose Methods)

Question to Answer	Preferred Method Combination	Key Data Requirements
Can it be done (Feasibility)	Stage-Gate Six Criteria + Policy & Compliance Check + Technology/Resource Availability + Pre-mortem	Policy texts, permits & resource evidence, similar failure cases
Is it worth doing (Economics)	Unit Economics Model + Financial Evaluation (3-scenario IRR/NPV/Payback/Break-even) + Sensitivity Critical Point + Bottom-up Market Sizing	Unit price & cost structure, investment composition, financing conditions, market sizing worksheet
Which option is better (Comparison)	Weighted Decision Matrix (or AHP approach) + Financial comparison per option + Conclusion robustness check	Quantifiable indicators per option; criteria weights and justification
How to make it happen (Path)	Stage-gate phased approach + Assumption ledger + Small-step validation (pilot/MVP)	Key assumptions list, affordable trial budget and time window

Method Library (Full version in references/方法论库.md, includes "When to use/How to use/What data needed/Common misuse")

1. Stage-Gate Decision (Go / Kill / Hold / Recycle + Six Criteria)
2. Bottom-Up Market Sizing (TAM / SAM / SOM)
3. Competitive Moat Analysis (7 Powers Structural Advantage Check)
4. Unit Economics Model (Unit Margin, Payback Period, LTV/CAC)
5. Financial Evaluation (IRR / NPV / Payback / Break-even / Sensitivity Critical Point)
6. Multi-Option Comparison (Weighted Decision Matrix / AHP Approach)
7. Pre-Mortem (Assume failure and trace back causes + early warning signals)
8. Assumption Ledger & Minimum Cost Validation
9. Dual-Perspective Cross (Policy View × Capital View)
10. Evidence Grading & Multi-Source Cross (A/B/C/D)

Selection Rule: 3–5 methods are enough for one pass; piling on methods ≠ rigor. Each method must be able to answer "what data does it need, what conclusion does it produce, and what if the conclusion is unsupported."

Step 4 · Execute Analysis: Six Mandatory Actions

4.1 Research First, Then Write (Order Enforced)

① Search & Due Diligence → ② Organize Data → ③ Cross-Validate → ④ Write. Never write first, then fill in data. Due diligence scope (parallel multi-track):

1. Industry Landscape: Scale, growth rate, lifecycle → Industry reports, brokerage research
2. Market: Target customer, demand intensity, willingness to pay → Competitors, user discussions, bid prices
3. Competition: Direct/indirect competitors, landscape, barriers → Top players
4. Policy: Industry policy, regulation, access, subsidies → Original government documents
5. Financials: Input-output benchmarks → Investment/return data of similar projects
6. Risk: Known failure cases, regulatory red lines → Search for negatives (This step is most often skipped, yet most valuable)

Escalate data collection based on page difficulty: Normal scraping → Anti-scraping bypass → Browser automation. Save raw materials for traceability; no need to re-search when revising.

4.2 Six Actions Checklist (Tailor to Project, Don't Force Fit)

#	Action	Core Question to Answer	Criterion
1	Market & Demand	How big is the market? Growth rate? Target audience? Why will we win?	Bottom-up estimation + ≥2 sources cross-verified
2	Competition & Barriers	Who are the competitors? What is our structural advantage?	At least one advantage that's not "work harder / be cheaper"; or honestly state "no barrier"
3	Unit Economics	Is each transaction profitable? How long to recoup customer acquisition cost?	Positive unit margin; payback period ≤ threshold value
4	Financial Evaluation	How much to invest? When to break even? IRR/NPV?	Three scenarios complete; critical points identified
5	Risk	Top 3 risks + trigger signals + countermeasures	Each risk has early warning indicators and an owner, not clichés
6	Compliance & Resources	Policy, permits, land, energy, qualifications	Status and estimated timeline for each item

4.3 Dual-Perspective Cross (Policy × Capital)

• Policy Perspective: Industry policy direction, access & approvals, regulatory trends, subsidies & compliance costs → Determines "Is now the right time?"
• Capital Perspective: Capital availability & cost, return thresholds, exit paths, comparable transaction valuation anchors → Determines "Is it worthwhile?"
• Cross Judgment: Policy support × insufficient capital = Small-step trial; Profitable capital × tightening policy = Cautious with exit plan; Both unsupportive = Don't do it.

Step 5 · Assumption Validation (The Most Valuable Step of This Skill)

The essence of project analysis is to list key assumptions, then validate them one by one.

5.1 List Assumptions, Then Prioritize

Write down all assumptions the conclusion depends on, and rank them by Impact (1-5) × Uncertainty (1-5) :

• High impact + High uncertainty → Validate first (This is "validate the most critical ones first, not the easiest ones")
• High impact + Low uncertainty → Low risk but must be traceable (if invalid, it's a game-changer)
• Low impact → Can be validated later

5.2 Each Assumption Must Have a "Pass/Fail Threshold"

Poor Wording	Good Wording
Customer is interested	Signed letters of intent from 3 regional top-tier customers (locking in 60% capacity)
Costs are controllable	Land acquisition cost ≤ 450,000 yuan/mu
Policy support	Included in provincial special plan / Obtained written confirmation from regulatory authorities
Technology feasible	Completed pilot run, stable operation for 30 consecutive days

If the threshold is not strict, validation becomes "finding evidence to prove oneself right."

5.3 Manage the Ledger with Scripts (Zero Dependencies)

# Assumption Ledger: Prioritization + Blockers + Validation Plan (shows what to validate first)
python3 scripts/assumption_ledger.py assumption_ledger.csv --plan
python3 scripts/assumption_ledger.py assumption_ledger.csv --json

# Feasibility Scorecard: Weighted score + Threshold judgment + Fatal assumptions + Go/Conditional Go/No-Go
python3 scripts/feasibility_scorecard.py project_evaluation.json
python3 scripts/feasibility_scorecard.py --print-schema   # See input structure
python3 scripts/feasibility_scorecard.py --demo           # See what output looks like

• Ledger CSV columns: Assumption, Category, Impact, Uncertainty, Validation Method, Evidence Level, Status, Owner, Deadline (Template: templates/assumption_ledger.csv)
• Status values: unverified / in_progress / verified / falsified
• Non-zero exit code = high-impact assumptions remain unverified, allowing integration into automation: no conclusion without passing.

5.4 Validation Methods Cost-Optimized (Don't Throw Money at It First)

Get primary data (policies, transaction announcements, tender prices) → Interview (customers, suppliers, industry peers) → Obtain pre-commitments (LOIs, credit letters, quotes) → Pilot/MVP (minimum viable scope) → Full commitment.

Step 6 · Conclusion & Decision Gate

Three Conclusion Formats (Don't just write "Feasible/Infeasible")

1. Go (Do it): All thresholds passed → Provide investment pace and next steps.
2. Conditional Go (Do it if): Primary thresholds met, blocking items exist → List "complete these X items to proceed" + validation plan and timeline.
3. No-Go (Don't do it): Falsified fatal assumption / Financial non-compliance / Red line crossed → Specify whether it's "direction is wrong" (Kill) or "conditions not met" (Hold).

Four Decision Gate Outputs (Source referenced at end): Go (Pass) / Kill (Terminate, document reasons) / Hold (Pause, specify signal to restart) / Recycle (Direction okay but insufficient material, send back for rework).

Practical Tip: For most "unclear" projects, the correct answer is Hold or Recycle, not a forced Go.

Conclusion Page Must Include: Overall judgment + Score table per dimension + Blocking items & fatal assumption status + Evidence level explanation + What to validate next (who, when, criterion).

Supporting File: templates/decision_gate_checklist.md (Six criteria + Threshold table + Four decisions)

Ten Evaluation Modules (Cut to Fit When Writing Reports)

#	Module	Core Question to Answer
1	Executive Summary	One-line conclusion + 3-5 core judgments (Write last)
2	Project Overview	What? Goal? Scope?
3	Background & Necessity	Why now? What if we don't?
4	Market Analysis	How big? Growth rate? Target audience? Why will we win?
5	Competition & Barriers	Who are the competitors? Our moat?
6	Product/Service & Operations	How will we deliver? Is the team sufficient?
7	Business Model	How will we make money? Revenue structure? Margin?
8	Financial Projections	How much to invest? When to break even? IRR/NPV?
9	Risk & Countermeasures	Top 3 risks + specific countermeasures + early warning signals
10	Conclusion & Recommendations	Go/No-Go/How + Next Steps

Writing Requirements

1. Write the Executive Summary last (you need conclusions first to write it accurately).
2. Every number must have a source or be marked "Assumption, pending validation."
3. Financial section: Provide calculation logic + key assumptions; do not fabricate numbers down to the cent.
4. Risk chapter: Write specific triggers + impact + countermeasures; no clichés.
5. Explicitly mark areas of uncertainty as "Pending validation / Pending completion" — acknowledging boundaries is professionalism, not a flaw.
6. Assumption ledger and conclusion page must be consistent with the main text (same number must have the same value in the summary, body, appendix, and conclusion page).
7. Delivery format per user request (Word/Markdown/PPT/online document), properly formatted, no "AI-like" feel.

Suggested Output Structure

• Cover Page (Title / Version / Date / Classification)
• Executive Summary (1 page): Overall judgment + Threshold table + Key assumption status + Next steps
• Body (Chapters trimmed from the ten modules)
• Appendix: Data Source Table (with evidence level), Key Assumptions Ledger, Pending Validation Checklist (who / when / criterion)

Quality Self-Checklist (Before Delivery)

• Are there any market numbers written without searching?
• Do core data points have ≥2 independent sources? Is the evidence level marked?
• Was the "what counts as a pass" threshold written down before starting, and does the conclusion map to each item?
• Do conclusion and evidence match? Is there any over-interpretation (e.g., correlation as causation, local as global)?
• Are opposing views / failure risks missing? Was a pre-mortem done?
• Do all key assumptions have a status (verified/unverified/falsified)? Are unverified high-impact assumptions explicitly listed as blocking items?
• Financials: Are the three scenarios complete? Are the four tables (Investment/Cost/Income/Indicators) consistent?
• Reader perspective: Can they make a decision after reading it?
• Has any "client's assumption" been written into the main text as "verified fact"?

Common Misuses & Traps

Trap	Consequence	Correction
Conclusion first, then find evidence	Analysis becomes advocacy	Threshold and assumption list are set before the analysis
Using the industry report's ceiling as your market	Severely overestimated size	Bottom-up estimation + differentiate TAM/SAM/SOM
Treating "work harder / be cheaper / good service" as a barrier	Project quickly commoditized	Use structural advantages (scale, network effects, switching costs)