A quick commercial reality check for offers that look good on the surface but may not hold up economically.

先交互，再分析

开始时先问清楚：

• 这次你想评估的 offer 是什么？

• 你们平时怎么判断一个 offer “可做”？

• 你希望沿用现有口径，还是让我给一套推荐的电商 profitability 框架？
• 这次最关心的是：利润、放量空间、转化假设，还是风险边界？

如果用户没有明确口径，先给推荐分析框架，再让用户确认。

Python script guidance

当用户提供结构化数字后：

• 生成 Python 脚本建模
• 先展示假设与公式
• 再输出 baseline / scenario / sensitivity
• 最后返回可复用脚本

如果关键数据缺失，不要直接假装精确；继续追问或给推荐默认值并等待确认。

Solves

Many ecommerce teams make pricing or offer decisions with incomplete economics:

• they see revenue upside but not margin drag;
• they model one variable but ignore knock-on effects;
• they test offers without clear guardrails;
• they scale offers before checking break-even logic.

Goal: Turn offer assumptions into a clearer economic view that is easier to evaluate and act on.

Use when

• You want to compare offer scenarios before launching
• A discount, bundle, or upsell idea sounds good but needs economic validation
• Growth teams need a faster way to pressure-test merchandising decisions
• Teams want clearer go / watch / no-go logic before scale

Inputs

• Core commercial assumptions relevant to the scenario
• Price and cost structure
• Margin or refund assumptions
• Traffic / conversion or attach-rate assumptions
• Constraints or guardrails

Workflow

• Clarify the baseline commercial setup and evaluation logic.
• Model the scenario inputs that change order economics.
• Surface upside, downside, and sensitivity.
• Identify the biggest weak points or break-even pressure.
• Recommend whether to test, revise, or avoid the scenario.
• Return reusable Python script when structured inputs exist.

Output

• Baseline view
• Scenario result
• Margin / break-even implications
• Key risks and weak points
• Recommendation
• Python script

Quality bar

• Output should be commercially interpretable, not just a raw formula dump.
• Recommendations should stay grounded in ecommerce economics.
• Weak points should be clearly separated from upside assumptions.
• The result should help a team decide what to test next.
• Do not pretend precision before assumptions are confirmed.

Resource

See references/output-template.md.