技能创建器
skill-creator
创建、改进和评估自定义技能,把专业流程转化为可重复使用的 AI 能力。
cost-benefit-analysis【数据与研究】Build a business case with costs, benefits, and ROI. Use when the user says "cost-benefit analysis", "business case", "ROI calculation", "is this worth it", "justify the investment", "payback period", "benefits realization", "financial justification", "total cost of ownership", "should we build or buy", "investment appraisal" - even if they don't exp…
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Based on The Business Analysis Handbook by Helen Winter - the Benefits Realization Framework that traces every requirement through output to benefit to business objective, ensuring nothing is built without a measurable business outcome. Also draws on Business Process Change by Paul Harmon for Activity-Based Costing in process improvements, and Business Analysis Techniques by James Cadle for the structured CBA technique with worked examples. The key insight from Winter: a cost-benefit analysis that only counts costs and tangible revenue misses the point. Benefits must be mapped to business objectives, and intangible benefits need proxy metrics.
State what is being evaluated:
DECISION: [What are we comparing?]
OPTIONS: [Build vs. Buy / Option A vs. B / Do vs. Don't]
TIME HORIZON: [3-5 years is standard]
DISCOUNT RATE: [organization's cost of capital or hurdle rate]
Categorize into one-time and recurring:
ONE-TIME COSTS:
- Development / procurement: $[amount]
- Data migration: $[amount]
- Training and change management: $[amount]
- Infrastructure setup: $[amount]
RECURRING COSTS (annual):
- Licensing / subscription: $[amount]
- Maintenance and support: $[amount]
- Operations and infrastructure: $[amount]
- Ongoing training: $[amount]
Use Harmon's Activity-Based Costing to trace costs to specific process activities rather than lumping into categories.
Trace each benefit to a business objective:
Requirement --> Output --> Benefit --> Business Objective
Example:
"Automated order validation" --> "Orders validated in < 1 min"
--> "Reduced processing cost by $50K/year" --> "Improve operational efficiency (BO-2)"
Classify benefits:
Build a year-by-year table:
| Year | Costs | Benefits | Net | Cumulative |
|------|-------|----------|-----|------------|
| 0 | -$[X] | $0 | -$[X] | -$[X] |
| 1 | -$[X] | +$[X] | +$[X] | -$[X] |
| 2 | -$[X] | +$[X] | +$[X] | +$[X] | <-- Payback
| ... | ... | ... | ... | ... |
Calculate:
For each major benefit and cost, assess:
ITEM: [benefit or cost]
BASE ESTIMATE: $[amount]
PROBABILITY: [% likelihood of realization]
RISK-ADJUSTED VALUE: $[base x probability]
KEY ASSUMPTION: [what must be true for this estimate to hold]
Run sensitivity analysis on the top 3 assumptions: what happens to NPV if each assumption is 20% worse?
For each option, present: total cost, total benefit, NPV, ROI, payback period, and key risks. Recommend the option with the best risk-adjusted NPV, not just the lowest cost or highest ROI.
# Cost-Benefit Analysis: [Decision Name]
## 1. Executive Summary (recommendation + key numbers)
## 2. Decision and Options
## 3. Cost Breakdown (one-time + recurring)
## 4. Benefits Map (Winter's chain: requirement -> output -> benefit -> objective)
## 5. Financial Summary (NPV, ROI, Payback table)
## 6. Risk and Sensitivity Analysis
## 7. Recommendation
## 8. Assumptions Register
1. Counting only tangible benefits Bad: CBA that shows $200K in development cost vs. $0 in quantified benefits because "it's hard to measure." Good: Use proxy metrics for intangibles: "Reduce manual effort by 20 hours/week = $52K/year at fully loaded cost."
2. Ignoring ongoing costs (TCO) Bad: Comparing $100K build cost to $80K buy cost (year 1 only). Good: Total Cost of Ownership over the full time horizon including maintenance, licensing, and operations.
3. Benefits without a realization owner Bad: "$500K in efficiency gains" with no one accountable for realizing them. Good: Each benefit has an owner who is responsible for tracking realization post-implementation.
4. No sensitivity analysis Bad: Single-point estimates presented as certain. Good: Show what happens if key assumptions change by +/- 20%. If NPV goes negative, the decision is fragile.
5. Comparing apples to oranges Bad: Comparing undiscounted future benefits to present-day costs. Good: All values discounted to present value using the same rate and time horizon.