原文核对：2026-10-13 · 中文正文 · 英文提示词与演示
场景定位
结合实验或上线计划、指标、队列和客户信号，量化提升、检查护栏并形成有来源的影响报告。
- 难度
- 中级
- 时间跨度
- 1 小时
适合用于
- 需要扩大、调整或停止建议的实验和项目。
- 需跨队列比较提升并检查护栏的团队。
- 需区分方法、局限、测量结果和解释的报告。
相关工具
起步提示词
英文原文 · 可复制后按实际材料修改
Measure whether [initiative, experiment, or launch] improved [target outcome].
Use the experiment or launch plan, success metrics, relevant dashboards, cohort or assignment data, customer signals, and launch notes I provide. Quantify lift or movement, check guardrail metrics, inspect segment differences, and explain whether the team should scale, change, or stop the initiative.
Return a business impact readout with charts, methodology notes, caveats, source links, and a clear recommendation. Separate confirmed results from interpretation and flag any missing inputs.从决策出发
明确扩大、调整或停止等决策，提供计划、成功指标、队列、护栏和客户背景，说明数据实际支持什么。
- 指定项目、目标、评估窗口和决策负责人。
- 提供计划、定义、分配/队列数据、仪表盘和上线笔记。
- 先验证输入并解释方法，再写建议。
- 检查提升、护栏和分群差异。
- 与分析师或实验负责人审阅方法、局限和缺失数据。
分开测量与解释；设计或数据不支持因果时收窄表述，不补造依据。
审计建议
分享前对建议和关键数字做简短证据审计。
官方提示词
英文原文 · 可复制后按实际材料修改
Audit the business impact readout.
Check:
- every material number against its source
- the lift calculation and comparison group
- guardrail metrics
- segment differences and sample gaps
- methodology limitations
- claims that go beyond the data
- whether scale, change, or stop is actually supported
Return corrections and open questions without changing the source data.展开英文原文对照
Give ChatGPT an experiment or launch plan, success metrics, cohort data, dashboard exports, customer signals, and launch notes, then ask it to quantify lift, check guardrails, explain segments, and draft a sourced impact readout.
Start with the decision the readout must support
An impact readout should connect the result to a decision, such as scale, adjust, or stop. Provide the experiment or launch plan, success metrics, cohorts, guardrails, and customer context so ChatGPT can show what the data supports.
- Name the initiative, target outcome, evaluation window, and decision owner.
- Attach the plan, metric definitions, assignment or cohort data, dashboards, and launch notes.
- Ask ChatGPT to validate inputs and explain the analysis method before writing the recommendation.
- Run the starter prompt and check lift, guardrails, and segment differences.
- Review methodology, caveats, and missing data with the analyst or experiment owner.
Keep measured results separate from interpretation. If the design or data cannot support a causal claim, ask ChatGPT to narrow the language rather than fill the gap.
Audit the recommendation
Before sharing the readout, ask for a compact evidence audit focused on the recommendation and its most important numbers.
官方提示词
英文原文 · 可复制后按实际材料修改
Audit the business impact readout.
Check:
- every material number against its source
- the lift calculation and comparison group
- guardrail metrics
- segment differences and sample gaps
- methodology limitations
- claims that go beyond the data
- whether scale, change, or stop is actually supported
Return corrections and open questions without changing the source data.