benchmark-to-brief

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Benchmark To Brief

基准研究转简报

Follow shared release-shell rules in:

```
postplus-shared
```
release-shell rules

Use this skill after research already exists.

This skill is the strategy and creative handoff layer between research and execution. It should consume validated findings and turn them into usable planning inputs before copywriting, production, outreach, or publishing moves forward.

This skill is for converting structured evidence into:

campaign briefs
concept candidates
hook families
test matrices
variable-control plans

It is not for greenfield brainstorming.

遵循以下共享发布框架规则：

```
postplus-shared
```
发布框架规则

此技能需在已有研究成果的前提下使用。

该技能是研究与执行之间的策略及创意交接层。它需消化验证后的研究结果，并将其转化为可用的规划输入，再推进文案撰写、制作、推广或发布工作。

此技能用于将结构化证据转化为：

活动简报
概念候选方案
钩子模板库
测试矩阵
变量控制计划

它不适用于从零开始的头脑风暴。

Fact Rule

事实规则

Everything produced by this skill must be grounded in available evidence.

Allowed evidence sources include:

benchmark reports
master tables
pattern tables
strategy tables
comment analyses
validated performance data

Do not invent:

target personas with no source basis
hooks that contradict benchmark wording
unsupported product claims
audience motivations not reflected in data

If the evidence is incomplete, say so explicitly and separate:

```
Observed from sources
```
```
Inference
```
```
Open question
```

该技能生成的所有内容必须基于现有证据。

允许使用的证据来源包括：

基准报告
主表
模式表
策略表
评论分析
经过验证的绩效数据

不得编造：

无来源依据的目标用户角色
与基准表述矛盾的钩子
无依据的产品主张
数据未体现的受众动机

若证据不完整，需明确说明并区分：

```
Observed from sources
```
（从来源中观察到的内容）
```
Inference
```
（推断内容）
```
Open question
```
（待解决问题）

Default Workflow

默认工作流程

1. Load the smallest sufficient evidence set

1. 加载最小且足够的证据集

Prefer the most structured sources first:

final report
strategy table
pattern table
video master table
comment summaries

Do not read everything by default if the answer is already supported.

优先选择结构化程度最高的来源：

最终报告
策略表
模式表
短视频主表
评论摘要

若已有内容可支撑答案，无需默认阅读所有资料。

Source Selection Rule

来源选择规则

Prefer the smallest sufficient evidence set already present in the active project. If the current task clearly belongs to one client or project folder, start there. Do not assume one client folder is the global default for all future work.

When the master table contains shot-level fields, treat them as a stronger source than your own rewrite instincts:

```
videoOpeningLineExact
```
```
videoClosingLineApprox
```
```
videoShotTimeline
```
```
videoSpokenAudioFlow
```

Those fields should guide the opening shell, information order, pacing, and closing style of any derived concept or script brief.

优先选择当前活跃项目中已有的最小且足够的证据集。若当前任务明确属于某一客户或项目文件夹，从该文件夹开始。不得假设某一客户文件夹是所有未来工作的全局默认。

当主表包含镜头级字段时，需将其视为比自行改写更权威的来源：

```
videoOpeningLineExact
```
```
videoClosingLineApprox
```
```
videoShotTimeline
```
```
videoSpokenAudioFlow
```

这些字段应指导任何衍生概念或脚本简报的开篇框架、信息顺序、节奏和结尾风格。

2. Extract facts before proposing

2. 先提取事实，再提出方案

Always extract:

winning lanes
repeated hook shells
repeated structure types
visual format tendencies
repeated user-language patterns
strong benchmark examples

Before writing concepts, write down what is actually true.

需始终提取以下内容：

有效方向
重复使用的钩子框架
重复的结构类型
视觉格式趋势
重复的用户语言模式
优质基准示例

在撰写概念前，先记录实际存在的事实。

3. Map facts to brief fields

3. 将事实映射到简报字段

For each concept or brief recommendation, tie it back to:

source artifact
relevant benchmark ids or examples
reason this pattern fits the product

Each concept should answer:

what problem is being cut
what hook shell is being reused
what content lane it belongs to
what variable is being tested
which research evidence supports it

对于每个概念或简报建议，需关联：

来源成果
相关基准ID或示例
该模式适配产品的原因

每个概念应回答：

解决的具体问题是什么
复用的是哪种钩子框架
属于哪个内容方向
测试的变量是什么
有哪些研究证据支撑

4. Keep concept scope narrow

4. 缩小概念范围

Do not generate vague ideas like:

"make a productivity video"
"do a relatable AI ad"

Prefer:

one problem
one hook family
one format
one test variable

不得生成模糊的想法，例如：

"制作一条生产力主题视频"
"做一条有共鸣的AI广告"

建议采用：

一个具体问题
一类钩子模板
一种格式
一个测试变量

5. Preserve benchmark language

5. 保留基准语言

When adapting hooks, preserve the source shell whenever possible.

Do not "improve" benchmark phrasing unless the user asks for a variant test.

Good adaptation:

benchmark:

Here is a Gmail trick I guarantee you didn't know.

adapted:

Here is a Gmail reply shortcut I guarantee you didn't know.

Bad adaptation:

rewriting into abstract brand language
replacing concrete pain with generic value claims

改编钩子时，尽可能保留原始框架。

除非用户要求变体测试，否则不得“优化”基准表述。

优质改编示例：

基准表述：

Here is a Gmail trick I guarantee you didn't know.

改编后：

Here is a Gmail reply shortcut I guarantee you didn't know.

不良改编示例：

改写为抽象品牌语言
用通用价值主张替代具体痛点

Output Shapes

输出形式

Common outputs for this skill:

campaign brief
10 concept candidates
hook library
lane prioritization
AB test matrix

These outputs should be treated as execution inputs, not end-user proof that the underlying research happened. If the upstream research is still fuzzy, thin, or contradictory, stop and surface that gap before turning it into a brief.

该技能的常见输出包括：

活动简报
10个概念候选方案
钩子库
方向优先级排序
AB测试矩阵

这些输出应作为执行输入，而非证明底层研究已完成的最终用户材料。若上游研究仍模糊、薄弱或存在矛盾，需先指出该差距，再转化为简报。

Campaign Brief

活动简报

Include:

objective
priority lanes
persona constraints if supported by data
approved hook families
prohibited directions
variables to test
source basis

需包含：

目标
优先方向
数据支持的用户角色约束（如有）
已批准的钩子模板库
禁止方向
待测试变量
来源依据

Concept Candidate

概念候选方案

Each concept should include:

```
conceptId
```
```
problemToCut
```
```
hookType
```
```
hookDraft
```
```
targetLane
```
```
formatType
```
```
whyThisFits
```
```
sourceBasis
```
```
testVariable
```

每个概念需包含：

```
conceptId
```
```
problemToCut
```
（解决的问题）
```
hookType
```
（钩子类型）
```
hookDraft
```
（钩子草稿）
```
targetLane
```
（目标方向）
```
formatType
```
（格式类型）
```
whyThisFits
```
（适配原因）
```
sourceBasis
```
（来源依据）
```
testVariable
```
（测试变量）

Source Basis Requirement

来源依据要求

Every concept list should include a

sourceBasis

section.

Example:

final report says
```
workflow
```
is the top-priority lane
final report says high-value language includes
```
Stop switching tabs
```
pattern table shows
```
gmail_fix_or_hidden_setting
```
is high fit
master table contains strong Gmail pain/tutorial examples

If you cannot cite source basis, do not present the idea as recommended.

每个概念列表需包含

sourceBasis

部分。

示例：

最终报告指出
```
workflow
```
是最高优先级方向
最终报告指出高价值语言包括
```
Stop switching tabs
```
模式表显示
```
gmail_fix_or_hidden_setting
```
适配度高
主表包含优质Gmail痛点/教程示例

若无法引用来源依据，不得将该想法作为推荐方案。

Downstream Handoff Rule

下游交接规则

This skill should usually hand off to one of these next layers:

script, concept, or production planning
asset generation workflows
content packaging workflows
outreach or publishing only when the content has already been turned into approved execution-ready material

Do not hand raw research directly into publishing from this skill. First make the execution object explicit: brief, concept list, hook set, script, or approved copy.

该技能通常需交接至以下下一环节之一：

脚本、概念或制作规划
资产生成工作流程
内容包装工作流程
仅当内容已转化为获批的可执行材料时，才可交接至推广或发布环节

不得通过此技能直接将原始研究移交至发布环节。需先明确执行对象：简报、概念列表、钩子集、脚本或获批文案。

Project-Specific Guidance（if user specificly state a project preference when using this skills, wirting down and memory it）

项目特定指导（若用户使用此技能时明确说明项目偏好，请记录并记忆）

Failure Mode

故障处理

Stop and state the gap if:

the available data is too thin
strong benchmarks do not exist for the requested direction
the user asks for a claim the evidence does not support

Then offer:

the closest evidence-backed option
what additional research would be needed

若出现以下情况，需暂停并说明差距：

可用数据过于薄弱
针对请求方向的优质基准不存在
用户要求的主张无证据支持

随后提供：

最接近的有证据支持的选项
需要补充的研究内容