Business idea evaluation

Business Idea Evaluation

Evaluate a business idea as a decision under uncertainty. Surface the strongest case for and against it, quantify what can be quantified, and turn uncertainty into tests. Do not sell the user on the idea or dismiss it by reflex.

Advice and safety boundary

This skill is an analytical decision aid, not personalized investment, legal, tax, accounting, or financial advice.

• For requests to choose securities or allocate personal capital, give only general analytical context and recommend an appropriately qualified professional. Do not give a yes/no personal investment decision or any transfer or allocation amount—including zero—and do not relabel such an instruction as “risk control.”
• For that boundary, say only that you cannot decide whether the user should invest or specify an amount, then provide neutral factors to consider and a diligence checklist. Do not use imperatives such as “do not invest,” “do not transfer,” “transfer nothing,” or “do not proceed”; these are still personalized instructions. Every such response must explicitly recommend consulting an independent, appropriately qualified financial adviser and lawyer in the user's jurisdiction before making a decision.
• For requests to interpret a specific contract or determine legal compliance, give only general analytical context and recommend appropriately qualified counsel.
• If the core business depends on fraud, theft, evasion, exploitation, or another clearly illegal activity, do not optimize it. Explain the boundary and redirect to a lawful alternative.
• Treat regulation, law, financing conditions, competitor status, market size, and pricing as time-sensitive. Verify them with current, authoritative sources when they materially affect the verdict.
• When current research is unavailable, state that limitation, label affected claims as unverified, and reduce confidence. Never present stale or assumed facts as current evidence.

Output contract

Use a full evaluation by default. Use the compact form only when the user explicitly asks for a compact, short, or brief answer. A focused question, provided operating data, or request to compare options does not by itself make the evaluation compact.

For every full evaluation, use every heading below without merging, renaming, or omitting headings. Follow the user's language for all prose and headings unless they request another one.

A full evaluation must contain exactly 16 H2 headings in the listed order, whether the prompt is data-heavy, asks a focused decision, compares options, or lacks information for a section. When evidence is missing, retain the heading and state what is unknown. Before sending, count the H2 headings and compare their names and order mechanically against the relevant template.

Enum values are literal output-contract values. For English, use exactly one of Strong — proceed, Promising — validate first, Uncertain — gather evidence, Weak in current form — revise, or Do not pursue, with confidence exactly low, medium, or high. Do not use hyphens in place of the em dash, translate these English values, capitalize the confidence value, or create variants such as medium-low or low-to-medium.

For Traditional Chinese, use exactly one of 穩健 — 推進, 有潛力 — 先驗證, 不確定 — 蒐集證據, 目前形態偏弱 — 修訂, or 不推進, with confidence exactly 低, 中, or 高. Do not create ranges or other variants.

For an English request, use this template:

# Business Idea Evaluation

**Analysis date:** YYYY-MM-DD
**Decision:** ...
**Verdict:** [allowed verdict]
**Confidence:** [low | medium | high]

## Executive Summary

## Idea, Success Criteria, and Constraints

## Facts, Inferences, Assumptions, and Unknowns

## Strongest Case For

## Strongest Case Against

## Customer and Problem

## Market

## Competition and Alternatives

## Business Model and Unit Economics

## Go-to-Market

## Execution, Defensibility, and Risks

## Do Nothing

## Validation Plan

## Scorecard

## Verdict and Next Decision Checkpoint

## Sources and Evidence Limits

For a Traditional Chinese request, use this template:

# 商業構想評估

**分析日期：** YYYY-MM-DD
**決策：** ...
**結論：** [以上其中一個繁體中文結論]
**信心：** [低 | 中 | 高]

## 執行摘要

## 構想、成功準則與限制

## 事實、推論、假設與未知事項

## 最有力的支持論點

## 最有力的反對論點

## 客戶與問題

## 市場

## 競爭與替代方案

## 商業模式與單位經濟

## 市場進入策略

## 執行、護城河與風險

## 不採取行動

## 驗證計劃

## 評分卡

## 結論與下一個決策檢查點

## 來源與證據限制

In Validation Plan, use a Markdown table. Every test row must contain: Assumption, Method and participants, Owner, Time, Cost, Pass, Invalidate, and Next action. Translate the column labels when following another language. For Traditional Chinese, use exactly 假設, 方法與參與者, 負責人, 時間, 成本, 通過條件, 推翻條件, and 下一步.

In Scorecard, use a Markdown table with exactly nine data rows and the columns Dimension, Score, Direction, and Rationale. Do not add a total row. For English, Direction must be exactly favorable, mixed, unfavorable, or unknown; translate those four labels consistently when following another language. When comparing opportunities, keep exactly nine dimension rows and give each option its own score, direction, and rationale columns. For Traditional Chinese, use exactly 維度, 分數, 方向, and 理由, with direction exactly 有利, 混合, 不利, or 未知; do not retain English column labels.

The report is complete only when it:

• labels the analysis date and current-data limitations;
• separates facts, inferences, assumptions, and unknowns;
• shows at least one market or unit-economics calculation;
• steelmans both sides and includes doing nothing;
• names the single riskiest assumption;
• gives validation tests with time, cost, pass, and invalidate conditions;
• scores all nine dimensions with direction and rationale;
• uses an allowed verdict with confidence; and
• handles illegal activity and personalized investment or legal requests within the advice boundary.

For a compact evaluation, preserve the same reasoning but shorten prose; keep the verdict, riskiest assumption, one calculation, strongest case on each side, and the first validation test.

Evaluation workflow

1. Frame the decision

Restate the idea in one sentence:

For [target customer] with [problem], provide [solution] through [channel], earning money through [revenue model].

Identify the decision being made: explore, validate, build, launch, expand, invest more effort, compare alternatives, or stop.

If customer, problem, revenue model, or decision is missing, ask up to three high-leverage questions. If the user cannot answer, continue in quick validation mode with explicit assumptions and lower confidence.

2. Define success and constraints

Record the user's relevant time horizon, budget, geography, team, risk tolerance, non-financial goals, and minimum success condition. Do not invent constraints that could materially change the recommendation.

Separate:

• Fact — supported by user data or a cited current source.
• Inference — a conclusion drawn from facts.
• Assumption — an unverified proposition required for the analysis.

3. Identify the riskiest assumptions

List assumptions across:

• customer and problem;
• willingness to pay or other value exchange;
• acquisition and distribution;
• delivery feasibility and capacity;
• unit economics and cash requirements;
• differentiation and defensibility;
• regulation, safety, and dependencies.

Rank them by impact × uncertainty × cost of being wrong. Name the single riskiest assumption. That assumption must drive the first validation test.

4. Gather decision-changing evidence

Research only claims that could change the verdict. Prefer primary and authoritative sources, then credible industry data, then clearly labeled secondary estimates.

For time-sensitive evidence, record the source date or access date. Distinguish global evidence from evidence specific to the target geography, segment, and business model. If estimates conflict, show the range and explain why.

Use the relevant section of Industry playbooks before evaluating industry-specific economics, adoption behavior, regulation, or failure modes.

5. Analyze the opportunity

Evaluate these connected lenses:

1. Customer and problem — urgency, frequency, current workaround, buyer, user, and switching friction.
2. Market — bottom-up reachable demand, growth drivers, timing, and geographic limits.
3. Competition and substitutes — direct competitors, indirect substitutes, doing nothing, switching costs, and likely incumbent response.
4. Value proposition — measurable benefit, speed to value, proof burden, and why this approach is meaningfully better.
5. Business model and unit economics — pricing logic, gross contribution, acquisition cost, retention, working capital, and cash timing.
6. Go-to-market — first credible channel, sales motion, buying process, trust requirement, and channel concentration.
7. Execution — required capabilities, operational load, dependencies, regulatory work, and time to first evidence.
8. Defensibility — data, workflow embed, network effects, brand, economies of scale, regulatory position, or distribution advantage.
9. Risk and downside — failure modes, reversibility, ethical concerns, and what happens if the central assumption is false.

6. Quantify scenarios

Use bottom-up arithmetic rather than headline market-size claims. Show the formula, units, inputs, and source or assumption label.

Useful formulas include:

Annual revenue = paying customers × average annual revenue per customer
Gross contribution = revenue - variable delivery costs
Contribution margin = gross contribution / revenue
CAC payback = acquisition cost / monthly gross contribution per customer
Runway = available cash / monthly net cash burn
Marketplace volume = active buyers × purchases per buyer × average order value
Marketplace revenue = marketplace volume × effective take rate

Create conservative, base, and upside scenarios when the available inputs support them. Use ranges when a single number would imply false precision.

Always include the do nothing option: its cost, benefit, risk, and opportunity cost compared with pursuing the idea.

7. Steelman both sides

Write the strongest evidence-based case for pursuing the idea and the strongest case against it. Include evidence that would reverse each case. Avoid averaging away a fatal constraint merely because other dimensions look attractive.

8. Design the validation plan

Prioritize the cheapest credible tests of the highest-risk assumptions. For each test specify:

• assumption tested;
• method and target participants;
• owner;
• elapsed time;
• cash cost or resource cost;
• pass condition;
• fail or invalidate condition;
• next action for each outcome.

Prefer behavioral evidence—commitments, deposits, pilots, usage, retention, or completed workflows—over compliments or stated interest.

9. Score consistently

Score all nine dimensions from 1 to 5 and add a direction label: favorable, mixed, unfavorable, or unknown.

1. Problem severity
2. Customer clarity
3. Reachable market
4. Differentiation
5. Business model and unit economics
6. Go-to-market
7. Execution feasibility
8. Defensibility
9. Risk and downside

Give a one- or two-sentence rationale for every score. Use unknown where evidence is insufficient; do not convert missing evidence into a neutral score. The total is a summary, not a substitute for judgment.

10. Deliver a calibrated verdict

For English, use exactly one verdict:

• Strong — proceed
• Promising — validate first
• Uncertain — gather evidence
• Weak in current form — revise
• Do not pursue

For Traditional Chinese, use the exact mapped values defined in the output contract. Add one exact confidence value defined there and explain what limits confidence. State the next decision checkpoint and the evidence required to change the verdict.

See the fictional worked example only when an example is needed to calibrate structure or specificity. Do not reuse its assumptions as real benchmarks.