Back to Details

planning-under-uncertainty

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Planning Under Uncertainty

不确定性下的规划

Scope

适用范围

Covers
  • Turning ambiguity into an executable plan via hypotheses, experiments, and decision triggers
  • Diagnosing “what’s actually happening” before acting (especially in crisis / wartime situations)
  • Using data as a compass (directional checks) rather than a GPS (false precision)
  • Building buffers and contingencies so the plan survives chaos
  • Setting a cadence for learning, decision-making, and stakeholder communication
When to use
  • “We need a plan, but the requirements are unclear and the outcome is uncertain.”
  • “Create a hypothesis-driven plan (experiments + decision rules) for this initiative.”
  • “We’re in a crisis (drop in retention/revenue/reliability) and need a wartime diagnosis + action plan.”
  • “Help us build contingencies, buffers, and pivot triggers before we commit.”
When NOT to use
  • You don’t agree on the underlying problem/opportunity (use 
    problem-definition
    ).
  • You need to choose what to do among many options (use 
    prioritizing-roadmap
    ).
  • You already have a clear plan and only need dates/milestones and stakeholder cadence (use 
    managing-timelines
    ).
  • You need a decision-ready PRD/spec for build execution (use 
    writing-prds
    / 
    writing-specs-designs
    ).
涵盖内容
  • 将模糊需求通过假设、实验和决策触发条件转化为可执行的计划
  • 在采取行动前诊断“实际情况”（尤其适用于危机/战时场景）
  • 将数据作为**指南针（方向校验）**而非GPS（虚假的精准性）
  • 建立缓冲机制与应急预案，确保计划能应对混乱
  • 设定节奏，用于学习、决策和利益相关者沟通
适用场景
  • “我们需要一个计划，但需求不明确，结果也不确定。”
  • “为这个项目创建一个基于假设的计划（包含实验和决策规则）。”
  • “我们正处于危机中（留存率/收入/可靠性下降），需要一套战时诊断+行动计划。”
  • “在我们投入资源前，帮我们制定应急预案、缓冲机制和转向触发条件。”
不适用场景
  • 你们对核心问题/机会未达成共识（使用
    problem-definition
    ）。
  • 你需要在多个选项中选择要做的事情（使用
    prioritizing-roadmap
    ）。
  • 你已经有清晰的计划，只需要确定日期/里程碑和利益相关者沟通节奏（使用
    managing-timelines
    ）。
  • 你需要一份可用于执行开发的PRD/规格文档（使用
    writing-prds
    / 
    writing-specs-designs
    ）。

Inputs

输入要求

Minimum required
  • The initiative context and desired outcome (“what are we trying to change?”)
  • Time horizon and urgency (wartime vs peacetime)
  • Constraints/guardrails (quality, compliance, brand, budget, “must not worsen” metrics)
  • Stakeholders and decision rights (who decides pivot/stop/scale?)
  • Top unknowns/assumptions (what would change the plan?)
  • Current signals (what data exists; what feels true but unproven?)
Missing-info strategy
  • Ask up to 5 questions from references/INTAKE.md.
  • If answers aren’t available, proceed with explicit assumptions and list Open questions that could change the plan.
最低必要输入
  • 项目背景和期望成果（“我们想要改变什么？”）
  • 时间范围和紧迫性（战时vs和平时期）
  • 约束条件/准则（质量、合规、品牌、预算、“不得恶化”的指标）
  • 利益相关者和决策权（谁决定转向/停止/扩大投入？）
  • 主要未知因素/假设（哪些情况会改变计划？）
  • 当前信号（已有的数据；哪些是感觉正确但未被证实的？）
缺失信息处理策略
  • 从[references/INTAKE.md]中最多提出5个问题。
  • 如果无法获得答案，基于明确的假设推进，并列出可能改变计划的未解决问题。

Outputs (deliverables)

输出成果（交付物）

Produce an Uncertainty Planning Pack in Markdown (in-chat; or as files if the user requests), containing:
  1. Decision frame (objective, “why now”, success + guardrails, time horizon, decision owner)
  2. Uncertainty map (assumptions/unknowns, confidence, impact, validation plan)
  3. Hypotheses + experiment portfolio (what we’ll learn, how, and what decision it enables)
  4. Plan v0 with buffers + contingencies (phases/options, triggers, fallbacks, pivot criteria)
  5. Cadence + comms (learning review ritual, update template, decision log)
  6. Risks / Open questions / Next steps (always included)
Templates: references/TEMPLATES.md
Expanded guidance: references/WORKFLOW.md
产出一份Markdown格式的不确定性规划包（可在对话中展示；若用户要求，也可作为文件输出），包含：
  1. 决策框架（目标、“为何此时启动”、成功标准+准则、时间范围、决策负责人）
  2. 不确定性地图（假设/未知因素、置信度、影响、验证计划）
  3. 假设与实验组合（我们要学习什么、如何学习、以及实验能支持哪些决策）
  4. 带缓冲与应急预案的V0版计划（阶段/选项、触发条件、 fallback方案、转向标准）
  5. 节奏与沟通方案（学习复盘机制、更新模板、决策日志）
  6. 风险/未解决问题/下一步行动（必须包含）
模板：[references/TEMPLATES.md]
扩展指南：[references/WORKFLOW.md]

Workflow (7 steps)

工作流程（7个步骤）

1) Intake + mode setting (wartime vs peacetime)

1) 信息收集+模式设定（战时vs和平时期）

  • Inputs: User request; references/INTAKE.md.
  • Actions: Clarify urgency, stakes, and what decision is needed. Decide whether you’re in diagnosis-first wartime mode or exploration peacetime mode.
  • Outputs: Short decision frame draft + mode declaration.
  • Checks: You can state: “We’re optimizing for <fast stabilization / learning / growth>. The decision we need by <date> is <pivot/stop/scale/commit>.”
  • 输入：用户需求；[references/INTAKE.md]。
  • 行动：明确紧迫性、风险级别以及所需做出的决策。确定采用先诊断的战时模式还是探索式的和平时期模式。
  • 输出：简短的决策框架草稿+模式声明。
  • 校验：你可以表述为：“我们正在优化<快速稳定/学习/增长>。我们需要在<日期>前做出<转向/停止/扩大投入/确认>的决策。”

2) Diagnose reality (humility first)

2) 诊断实际情况（保持谦逊）

  • Inputs: Current signals, anecdotes, dashboards, incident reports, qualitative inputs.
  • Actions: Separate symptoms from hypotheses. Write 3–7 plausible explanations, and identify what evidence would falsify each. Avoid prematurely picking a favorite story.
  • Outputs: “What we know / don’t know” + initial hypothesis set.
  • Checks: At least one hypothesis contradicts the team’s initial intuition (to reduce confirmation bias).
  • 输入：当前信号、轶事、仪表盘、事件报告、定性输入。
  • 行动：区分症状与假设。撰写3-7个合理的解释，并确定能推翻每个解释的证据。避免过早偏向某一种说法。
  • 输出：“我们已知/未知的内容”+初始假设集。
  • 校验：至少有一个假设与团队的初始直觉相反（以减少确认偏差）。

3) Build the uncertainty map (assumptions → validation plan)

3) 构建不确定性地图（假设→验证计划）

  • Inputs: Hypotheses; constraints; stakeholders; time horizon.
  • Actions: Create an uncertainty map of assumptions/unknowns with confidence and impact; prioritize the top items that would change the plan.
  • Outputs: Uncertainty map table + prioritized “top 5 unknowns”.
  • Checks: Every top unknown has a clear validation method and an owner.
  • 输入：假设集；约束条件；利益相关者；时间范围。
  • 行动：创建包含假设/未知因素、置信度和影响的不确定性地图；优先列出会改变计划的核心未知因素。
  • 输出：不确定性地图表格+优先级最高的“5大未知因素”。
  • 校验：每个核心未知因素都有明确的验证方法和负责人。

4) Define hypotheses + decision rules (learning over “wins”)

4) 定义假设+决策规则（学习优先于“成功”）

  • Inputs: Top unknowns; success/guardrails; risk tolerance.
  • Actions: Turn unknowns into testable hypotheses. For each hypothesis, define: expected learning, success signal(s), guardrails, and the decision the result enables (stop/pivot/scale).
  • Outputs: Hypothesis statements + decision rules.
  • Checks: Each hypothesis ties to a decision; “winning” is defined as learning, not just positive results.
  • 输入：核心未知因素；成功标准/准则；风险容忍度。
  • 行动：将未知因素转化为可测试的假设。针对每个假设，定义：预期学习成果、成功信号、准则、以及实验结果能支持的决策（停止/转向/扩大投入）。
  • 输出：假设陈述+决策规则。
  • 校验：每个假设都与决策挂钩；“成功”的定义是获得学习成果，而非仅得到积极结果。

5) Design a reproducible testing process (many shots at bat)

5) 设计可重复的测试流程（多次尝试）

  • Inputs: Hypothesis set; available tools; team capacity.
  • Actions: Create an experiment portfolio that balances speed vs confidence (smoke tests, prototypes, A/Bs, customer calls, operational drills). Set a cadence to run and review tests continuously.
  • Outputs: Experiment portfolio table + review cadence.
  • Checks: At least 1 fast test can run within the next 1–2 weeks (or faster in wartime).
  • 输入：假设集；可用工具；团队能力。
  • 行动：创建平衡速度与置信度的实验组合（冒烟测试、原型、A/B测试、客户访谈、运营演练）。设定持续运行和复盘测试的节奏。
  • 输出：实验组合表格+复盘节奏。
  • 校验：至少有1个快速测试能在未来1-2周内启动（战时场景需更快）。

6) Turn learning into a plan with buffers, contingencies, and triggers

6) 将学习成果转化为带缓冲、应急方案和触发条件的计划

  • Inputs: Experiment portfolio; constraints; dependencies; timeline needs.
  • Actions: Draft Plan v0 with phases/options; add buffers; define contingencies and explicit triggers for pivot/rollback/escalation. Use data as a compass: focus on directional signals and early warnings, not false certainty.
  • Outputs: Plan v0 + buffer/contingency section + trigger list.
  • Checks: There is a clear “if X happens, we will do Y” for the top risks/unknowns.
  • 输入：实验组合；约束条件；依赖关系；时间线需求。
  • 行动：起草V0版计划，包含阶段/选项；添加缓冲机制；定义应急预案和转向/回滚/升级的明确触发条件。将数据作为指南针：关注方向信号和早期预警，而非虚假的确定性。
  • 输出：V0版计划+缓冲/应急方案部分+触发条件列表。
  • 校验：针对核心风险/未知因素，存在明确的“如果X发生，我们将执行Y”的规则。

7) Quality gate + finalize

7) 质量校验+最终定稿

  • Inputs: Full draft pack.
  • Actions: Run references/CHECKLISTS.md and score with references/RUBRIC.md. Ensure Risks / Open questions / Next steps exist with owners and time bounds.
  • Outputs: Final Uncertainty Planning Pack.
  • Checks: A stakeholder can approve the plan async and the team can execute without re-litigating the ambiguity.
  • 输入：完整的规划包草稿。
  • 行动：使用[references/CHECKLISTS.md]进行检查，并通过[references/RUBRIC.md]进行评分。确保风险/未解决问题/下一步行动部分明确包含负责人和时间期限。
  • 输出：最终的不确定性规划包。
  • 校验：利益相关者可异步批准该计划，团队无需再纠结模糊点即可执行。

Quality gate (required)

质量校验（必填）

  • Use references/CHECKLISTS.md and references/RUBRIC.md.
  • Always include: Risks, Open questions, Next steps.
  • 使用[references/CHECKLISTS.md]和[references/RUBRIC.md]。
  • 必须包含：风险、未解决问题、下一步行动。

Examples

示例

Example 1 (ambiguous initiative): “We think onboarding is hurting conversion, but we’re not sure why. Create an uncertainty plan with hypotheses, experiments, and pivot triggers.”
Expected: an uncertainty map + experiment portfolio (qual + quant) + a Plan v0 that commits to learning milestones, not premature delivery dates.
Example 2 (wartime): “Retention dropped 15% this week after a release. We need a wartime plan: diagnose root causes, run rapid tests, and decide whether to rollback or patch.”
Expected: diagnosis-first workflow with falsifiable hypotheses, tight guardrails, and explicit rollback/escalation triggers.
Boundary example: “Write a full PRD for Feature X.”
Response: clarify uncertainty first (this skill), then use 
writing-prds
once the hypotheses, constraints, and decision gates are clear.
示例1（模糊项目）：“我们认为新用户注册流程影响了转化率，但不确定原因。创建一个包含假设、实验和转向触发条件的不确定性规划。”
预期产出：不确定性地图+实验组合（定性+定量）+V0版计划，该计划承诺学习里程碑，而非过早设定交付日期。
示例2（战时场景）：“本周发布新版本后，留存率下降了15%。我们需要一套战时计划：诊断根本原因、开展快速测试、并决定是回滚还是修复。”
预期产出：先诊断的工作流程，包含可证伪的假设、严格的准则、以及明确的回滚/升级触发条件。
边界示例：“为功能X撰写完整的PRD。”
回应：先通过本技能明确不确定性，待假设、约束条件和决策节点清晰后，再使用
writing-prds
。