icp-website-review

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

ICP Website Review

ICP 网站评估

Evaluate any online asset through the eyes of pre-built synthetic ICP personas. This is the evaluation engine — it loads personas created by

icp-persona-builder

and runs them against whatever you point it at.

通过预构建的合成ICP personas视角评估任何在线资产。这是评估引擎——它加载由

icp-persona-builder

创建的画像，并针对你指定的任何对象运行评估。

Prerequisites

前提条件

Personas must exist at

clients/<client>/personas/personas.json

. If they don't, run

icp-persona-builder

first.

Personas必须存储在

clients/<client>/personas/personas.json

路径下。如果不存在，请先运行

icp-persona-builder

。

Quick Start

快速开始

Structured scorecard (default):

Review [company]'s website using their ICP personas. Site: [url].

Freeform focus group:

Run a focus group on [company]'s new landing page: [url]. Use focus-group mode.

Head-to-head competitive:

Compare [company]'s site ([url]) against [competitor] ([url]) through ICP personas.

结构化评分卡（默认模式）：

Review [company]'s website using their ICP personas. Site: [url].

自由形式焦点小组：

Run a focus group on [company]'s new landing page: [url]. Use focus-group mode.

一对一竞品对比：

Compare [company]'s site ([url]) against [competitor] ([url]) through ICP personas.

Inputs

输入参数

Input	Required	Source
Personas	Yes	`clients/<client>/personas/personas.json`
Target URL(s)	Yes	User provides
Mode	No	`scorecard` (default), `focus-group` , or `head-to-head`
Competitor URL(s)	For head-to-head	User provides
Specific pages	No	Default: homepage + key pages discovered during crawl
Scope	No	`full-site` (default) or specific page URL for a single-page review

输入项	是否必填	来源
Personas	是	`clients/<client>/personas/personas.json`
目标URL	是	用户提供
模式	否	`scorecard` （默认）、 `focus-group` 或 `head-to-head`
竞品URL	仅一对一模式需要	用户提供
特定页面	否	默认：首页 + 爬取过程中发现的关键页面
范围	否	`full-site` （默认）或单页评估的特定页面URL

Modes

模式说明

Mode 1: Structured Scorecard (Default)

模式1：结构化评分卡（默认）

Rigorous, dimension-by-dimension evaluation with numerical scores. Best for:

Pre-redesign audits
Conversion optimization
Tracking improvement over time (comparable scores across runs)
Stakeholder presentations (data-driven)

严格的维度化评估，附带数值评分。最适用于：

网站重设计前的审计
转化优化
跟踪随时间的改进（跨评估周期的可对比评分）
向利益相关者汇报（数据驱动）

Mode 2: Freeform Focus Group

模式2：自由形式焦点小组

Each persona reacts naturally — stream of consciousness, emotional reactions, questions they'd have, things that confuse them. No rigid scoring. Best for:

Early-stage feedback on a draft or concept
Understanding emotional reactions and gut feelings
Surfacing unexpected issues the scorecard dimensions might miss
More natural, human-sounding feedback

每个Personas做出自然反应——意识流、情绪反馈、他们会提出的问题、感到困惑的点。无严格评分。最适用于：

草稿或概念的早期反馈
了解情绪反应和直觉感受
发现评分卡维度可能遗漏的意外问题
更自然、更贴近真实用户的反馈

Mode 3: Head-to-Head Competitive

模式3：一对一竞品对比

Each persona evaluates your site AND a competitor's site, then picks a winner with reasoning. Best for:

Competitive positioning
Understanding why prospects choose competitors
Finding specific areas where competitors outperform you
Sales battlecard input

每个Personas同时评估你的网站和竞品网站，然后给出选择结果及理由。最适用于：

竞争定位
理解潜在客户选择竞品的原因
发现竞品表现更优的具体领域
销售作战卡片的输入素材

Process

流程

Step 1: Load Personas

步骤1：加载Personas

Read

clients/<client>/personas/personas.json

. Confirm:

Personas exist and are well-formed
List the personas that will be used (name, title, segment)
If the user wants to use only specific personas, filter accordingly

读取

clients/<client>/personas/personas.json

文件。确认：

Personas存在且格式规范
列出将使用的Personas（姓名、职位、细分群体）
如果用户只想使用特定Personas，进行相应过滤

Step 2: Crawl Target Pages

步骤2：爬取目标页面

Fetch the key pages each persona would realistically visit:

Homepage (everyone starts here)
Pricing page (if public)
Product/features page (technical buyers)
Solutions/use-cases page (business buyers)
About page (trust/credibility check)
Case studies or testimonials (social proof)
Blog (1-2 recent posts, for content quality signal)
Documentation or resources (if relevant for technical buyers)

Use

WebFetch

for each page. Extract main content, headlines, CTAs, and overall structure. Note what's present and what's missing.

Also check external presence:

Search results: WebSearch "[company name]" and "[company name] reviews"
Review sites: Quick G2/Capterra check
Social proof signals: Notable logos, press, trust badges

For head-to-head mode, crawl the same pages on the competitor site.

获取每个Personas实际会访问的关键页面：

首页（所有人的起始页）
定价页（如果公开）
产品/功能页（技术买家）
解决方案/用例页（业务买家）
关于我们页（信任/可信度验证）
案例研究或客户证言（社交证明）
博客（1-2篇近期文章，用于内容质量信号）
文档或资源页（如果与技术买家相关）

对每个页面使用

WebFetch

工具。提取主要内容、标题、CTA（行动召唤）和整体结构。记录存在的内容和缺失的内容。

同时检查外部曝光情况：

搜索结果：使用WebSearch搜索“[公司名称]”和“[公司名称] 评价”
评价网站：快速查看G2/Capterra
社交证明信号：知名客户标志、媒体报道、信任徽章

对于一对一模式，爬取竞品网站的相同页面。

Step 3: Run Evaluations (Mode-Dependent)

步骤3：运行评估（依模式而定）

Scorecard Mode

评分卡模式

For each persona, evaluate on these dimensions:

A) First Impression (10 seconds)

What do they think this company does?
Is it clear this is for them?
Does the headline speak to their pain point?
Trust level at first glance?
Rating: 1-10

B) Messaging Relevance

Does the value proposition resonate with their specific pain?
Is the language familiar or alienating?
Do they see themselves in the messaging?
Are benefits framed in terms they care about?
Rating: 1-10

C) Trust & Credibility

Social proof relevant to them? (Same industry, size, role)
Claims backed by evidence?
Team/about page build confidence?
Any red flags?
Rating: 1-10

D) Clarity & Information Architecture

Can they find what they need quickly?
Is the product easy to understand?
Is pricing clear?
Are next steps obvious?
Rating: 1-10

E) Objection Handling

Does the site address their likely objections?
Comparison pages or differentiators available?
Enough depth for their decision process?
Rating: 1-10

F) Overall Verdict

Would they take the next step?
What's the #1 blocker?
Single most impactful improvement?
Overall score: 1-10

针对每个Personas，从以下维度进行评估：

A) 第一印象（10秒内）

他们认为这家公司是做什么的？
是否能明确感知到这是针对他们的产品/服务？
标题是否触达了他们的痛点？
第一眼的信任度如何？
评分：1-10

B) 信息相关性

价值主张是否与他们的特定痛点产生共鸣？
使用的语言是熟悉的还是令人疏远的？
他们是否能从信息中找到自身的定位？
利益点是否以他们关心的方式呈现？
评分：1-10

C) 信任与可信度

社交证明是否与他们相关？（同行业、同规模、同职位）
主张是否有证据支持？
团队/关于我们页是否能建立信心？
是否存在任何危险信号？
评分：1-10

D) 清晰度与信息架构

他们能否快速找到所需内容？
产品是否易于理解？
定价是否清晰？
下一步行动是否明确？
评分：1-10

E) 异议处理

网站是否解决了他们可能提出的异议？
是否有对比页面或差异化说明？
内容深度是否满足他们的决策流程？
评分：1-10

F) 总体结论

他们会采取下一步行动吗？
最大的障碍是什么？
最具影响力的改进点是什么？
总体评分：1-10

Focus Group Mode

焦点小组模式

For each persona, produce a natural, first-person reaction. No rigid structure — let the persona talk.

Prompt each persona with:

You are [persona name], [title] at [company type]. [Situation summary]. You just landed on this website. React naturally — what do you notice first? What confuses you? What excites you? What makes you skeptical? Would you keep exploring or bounce? If you stayed, where would you go next? What questions do you need answered before you'd take action?

Guidelines:

Write in first person as the persona
Let their personality and skepticism level come through
Include emotional reactions ("this immediately makes me nervous because...")
Reference specific things on the page ("the headline says X but I was looking for Y")
Note what they'd do next ("I'd probably Google '[competitor] vs [company]' after seeing this")
End with a gut-level verdict: interested, skeptical, confused, or gone

After all personas react, synthesize common themes and divergent opinions.

针对每个Personas，生成自然的第一人称反应。无严格结构——让Personas自由表达。

给每个Personas的提示：

你是[Personas姓名]，[公司类型]的[职位]。[场景概述]。你刚刚进入这个网站。自然地做出反应——你首先注意到什么？什么让你困惑？什么让你兴奋？什么让你怀疑？你会继续浏览还是离开？如果留下，你接下来会去哪里？在采取行动前，你需要哪些问题的答案？

指导原则：

以Personas的第一人称写作
体现他们的个性和怀疑程度
包含情绪反应（“这立刻让我紧张，因为……”）
引用页面上的具体内容（“标题说X，但我要找的是Y”）
记录他们接下来会做什么（“看到这个后，我可能会搜索‘[竞品] vs [公司]’”）
以直觉结论结尾：感兴趣、怀疑、困惑或离开

所有Personas反应完成后，总结共同主题和不同观点。

Head-to-Head Mode

一对一竞品对比模式

For each persona, evaluate BOTH sites, then compare.

Structure per persona:

Quick take on Site A (3-5 bullet reactions)
Quick take on Site B (3-5 bullet reactions)
Dimension-by-dimension comparison:

Dimension	Site A	Site B	Winner
First Impression	[X/10]	[X/10]	[A/B/Tie]
Messaging Relevance	[X/10]	[X/10]	[A/B/Tie]
Trust & Credibility	[X/10]	[X/10]	[A/B/Tie]
Clarity & Navigation	[X/10]	[X/10]	[A/B/Tie]
Objection Handling	[X/10]	[X/10]	[A/B/Tie]
Overall	[X/10]	[X/10]	[A/B/Tie]

"If I had to choose today..." — Which site wins and why, in this persona's own words
What Site A could steal from Site B — Specific things the competitor does better
What Site A does better — Don't lose these advantages

After all personas, produce a cross-persona competitive summary:

Where does each site win across most personas?
Which persona segments lean toward the competitor? (These are at-risk segments)
What are the competitor's key advantages to neutralize?
What are your unique advantages to amplify?

针对每个Personas，同时评估两个网站，然后进行对比。

每个Personas的结构：

对网站A的快速评价（3-5条要点）
对网站B的快速评价（3-5条要点）
维度化对比：

维度	网站A	网站B	胜出方
第一印象	[X/10]	[X/10]	[A/B/平局]
信息相关性	[X/10]	[X/10]	[A/B/平局]
信任与可信度	[X/10]	[X/10]	[A/B/平局]
清晰度与导航	[X/10]	[X/10]	[A/B/平局]
异议处理	[X/10]	[X/10]	[A/B/平局]
总体	[X/10]	[X/10]	[A/B/平局]

“如果今天必须选择……” —— 以Personas的口吻说明选择哪个网站及原因
网站A可以从网站B借鉴的点 —— 竞品做得更好的具体方面
网站A的优势 —— 不要丢失这些优势

所有Personas评估完成后，生成跨Personas的竞品总结：

哪些维度下，多数Personas认为某网站胜出？
哪些Personas细分群体倾向于竞品？（这些是高风险群体）
竞品的核心优势是什么，需要如何抵消？
你的独特优势是什么，需要如何强化？

Step 4: Cross-Persona Synthesis

步骤4：跨Personas总结

Regardless of mode, always produce:

Consensus issues — Flagged by ALL personas (urgent)
Segment-specific gaps — Only certain personas noticed (targeted fixes)
Messaging disconnects — Site language vs. buyer language
Missing content — Pages, sections, or proof points personas wanted
Strengths to preserve — Don't break what works
Priority matrix — Ranked by impact (personas affected) and effort

无论使用哪种模式，都必须生成以下内容：

共识问题 —— 所有Personas都指出的问题（紧急）
细分群体特定差距 —— 仅部分Personas注意到的问题（针对性修复）
信息脱节 —— 网站语言与买家语言的差异
缺失内容 —— Personas期望的页面、板块或证明点
需保留的优势 —— 不要破坏有效的部分
优先级矩阵 —— 按影响范围（涉及的Personas数量）和实施难度排序

Output

输出

Save to

clients/<client-name>/icp-reviews/<date>-<mode>.md

This path allows multiple reviews over time — tracking improvements, comparing before/after redesigns, or accumulating competitive intel.

保存至

clients/<client-name>/icp-reviews/<date>-<mode>.md

该路径支持随时间进行多次评估——跟踪改进情况、对比重设计前后的效果，或积累竞品情报。

Scorecard Output Template

评分卡输出模板

markdown

undefined

markdown

undefined

ICP Website Review: [Company Name]

Date: [Date] | Mode: Scorecard | URL: [website] Pages analyzed: [N] | Personas used: [N] Personas from:

clients/<client>/personas/personas.json

(created [date])

Date: [Date] | Mode: Scorecard | URL: [website] Pages analyzed: [N] | Personas used: [N] Personas from:

clients/<client>/personas/personas.json

(created [date])

Executive Summary

[3-5 sentences: overall assessment, biggest strengths, biggest gaps, top 3 recs]

Average score: [X/10]

Persona	Segment	Score	Would Convert?
[Name]	[Segment]	[X/10]	[Yes/Maybe/No]
...	...	...	...

[3-5 sentences: overall assessment, biggest strengths, biggest gaps, top 3 recs]

Average score: [X/10]

Persona	Segment	Score	Would Convert?
[Name]	[Segment]	[X/10]	[Yes/Maybe/No]
...	...	...	...

Persona Reviews

[Persona Name]'s Review

Arriving with: [Their situation and buying trigger]

Dimension	Score	Summary
First Impression	[X/10]	[One line]
Messaging Relevance	[X/10]	[One line]
Trust & Credibility	[X/10]	[One line]
Clarity & Navigation	[X/10]	[One line]
Objection Handling	[X/10]	[One line]
Overall	[X/10]	[One line]

Liked:

[Specific, with reference from the site]

Frustrated by:

[Specific, with explanation of why THIS persona cares]

Wished they could find:

[Missing content/info, why they need it]

Verdict: [Would they convert? #1 blocker?]

[Repeat for each persona]

Arriving with: [Their situation and buying trigger]

Dimension	Score	Summary
First Impression	[X/10]	[One line]
Messaging Relevance	[X/10]	[One line]
Trust & Credibility	[X/10]	[One line]
Clarity & Navigation	[X/10]	[One line]
Objection Handling	[X/10]	[One line]
Overall	[X/10]	[One line]

Liked:

[Specific, with reference from the site]

Frustrated by:

[Specific, with explanation of why THIS persona cares]

Wished they could find:

[Missing content/info, why they need it]

Verdict: [Would they convert? #1 blocker?]

[Repeat for each persona]

Cross-Persona Analysis

Consensus Issues

[Issue] — [Why it matters]

[Issue] — [Why it matters]

Segment-Specific Gaps

Gap	Affected Personas	Impact
[Gap]	[Names]	[High/Med/Low]

Gap	Affected Personas	Impact
[Gap]	[Names]	[High/Med/Low]

Messaging Disconnects

Site says	Buyers say	Affected
"[quote]"	"[quote]"	[Names]

Site says	Buyers say	Affected
"[quote]"	"[quote]"	[Names]

Strengths to Preserve

[What works]

[What works]

Prioritized Recommendations

High Impact

[Rec] — Affects: [personas]. Current: [X]. Target: [Y].

[Rec] — Affects: [personas]. Current: [X]. Target: [Y].

Medium Impact

[Rec]

[Rec]

Quick Wins

[Small changes]

[Small changes]

Score Matrix

Dimension	[P1]	[P2]	[P3]	[P4]	Avg
First Impression
Messaging
Trust
Clarity
Objections
Overall

undefined

Dimension	[P1]	[P2]	[P3]	[P4]	Avg
First Impression
Messaging
Trust
Clarity
Objections
Overall

undefined

Focus Group Output Template

焦点小组输出模板

markdown

undefined

markdown

undefined

ICP Focus Group: [Company Name]

Date: [Date] | Mode: Focus Group | URL: [website] Personas used: [N]

Executive Summary

[What the group collectively thought — 3-5 sentences]

Persona Reactions

[Persona Name] — [Title]

[First-person narrative reaction, 200-400 words. Natural voice, emotional, specific.]

Gut verdict: [One sentence — interested/skeptical/confused/gone]

[Repeat for each persona]

[First-person narrative reaction, 200-400 words. Natural voice, emotional, specific.]

Gut verdict: [One sentence — interested/skeptical/confused/gone]

[Repeat for each persona]

Group Synthesis

What everyone noticed

[Theme]

[Theme]

Where opinions split

Topic	[P1] take	[P2] take

Topic	[P1] take	[P2] take

Strongest reactions (positive)

[What got the most enthusiasm, from whom]

[What got the most enthusiasm, from whom]

Strongest reactions (negative)

[What got the most criticism, from whom]

[What got the most criticism, from whom]

What nobody could answer

[Questions the site left unanswered]

[Questions the site left unanswered]

Recommendations

[Prioritized, grouped by theme]

undefined

[Prioritized, grouped by theme]

undefined

Head-to-Head Output Template

一对一竞品对比输出模板

markdown

undefined

markdown

undefined

Competitive Review: [Company] vs [Competitor]

Date: [Date] | Mode: Head-to-Head Site A: [url] | Site B: [url] Personas used: [N]

Executive Summary

Overall winner: [A/B/Split] [3-5 sentences on where each site wins and loses]

Persona	Segment	Site A	Site B	Pick
[Name]	[Segment]	[X/10]	[X/10]	[A/B]
...	...	...	...	...

Overall winner: [A/B/Split] [3-5 sentences on where each site wins and loses]

Persona	Segment	Site A	Site B	Pick
[Name]	[Segment]	[X/10]	[X/10]	[A/B]
...	...	...	...	...

Persona Comparisons

[Persona Name]

Site A quick take: [3-5 bullets] Site B quick take: [3-5 bullets]

Dimension	A	B	Winner
...

"If I had to choose..." [Persona's verdict in their own words]

What A should steal from B: [Specific] What A does better: [Specific]

[Repeat for each persona]

Site A quick take: [3-5 bullets] Site B quick take: [3-5 bullets]

Dimension	A	B	Winner
...

"If I had to choose..." [Persona's verdict in their own words]

What A should steal from B: [Specific] What A does better: [Specific]

[Repeat for each persona]

Competitive Summary

Where [Company] wins across personas

[Advantage, which personas care]

[Advantage, which personas care]

Where [Competitor] wins across personas

[Advantage, which personas care]

[Advantage, which personas care]

At-risk segments

[Personas who lean toward competitor — these ICPs may be churning or not converting]

Priority moves

Neutralize: [Competitor advantage to close the gap on]
Amplify: [Your advantage to lean into harder]
Differentiate: [Unique angle neither site owns yet]

undefined

Neutralize: [Competitor advantage to close the gap on]
Amplify: [Your advantage to lean into harder]
Differentiate: [Unique angle neither site owns yet]

undefined

Running Repeat Reviews

重复运行评估

Since personas are saved separately, you can re-run this skill anytime:

Re-run the ICP website review for [client]. Use the existing personas.

Compare results across dates to track improvement:

Compare today's ICP review for [client] against the one from [date].

The dated output path (

icp-reviews/<date>-<mode>.md

) makes it easy to see the history of reviews and how scores change over time.

由于Personas单独存储，你可以随时重新运行此功能：

Re-run the ICP website review for [client]. Use the existing personas.

对比不同日期的结果以跟踪改进：

Compare today's ICP review for [client] against the one from [date].

带日期的输出路径（

icp-reviews/<date>-<mode>.md

）便于查看评估历史，以及分数随时间的变化情况。

Tips

提示

Don't be nice. Personas should be honest and critical. A persona that says "everything looks great" is useless.
Vary the voices. A CTO's review should read differently than a marketing manager's. Use different vocabulary, different focus, different emotional register.
Ground in specifics. Not "messaging is unclear" but "the headline 'Accelerate Your Workflow' doesn't tell me what this product does — I had to scroll to the third fold."
Check the competition. If personas mention alternatives, quickly check 1-2 competitor sites to see if they handle the same issues better.
External presence matters. What shows up when a persona Googles the company? Reviews? Press? That shapes trust before they even hit the site.
Focus group mode is underrated. The freeform reactions often surface things the scorecard misses — emotional reactions, unexpected confusion, things people wouldn't flag in a survey.
Head-to-head is the killer use case. Nothing clarifies your positioning gaps faster than seeing your site through a buyer's eyes right next to the competitor's.
This skill has no code script. It's agent-executed using WebSearch and WebFetch.

不要留情面。Personas应诚实且挑剔。一个说“一切看起来都很好”的Personas是毫无用处的。
多样化语气。CTO的评估应与营销经理的评估风格不同。使用不同的词汇、不同的关注点、不同的情绪基调。
基于具体细节。不要说“信息不清晰”，而是说“标题‘加速你的工作流’没有告诉我这个产品是做什么的——我不得不滚动到第三屏才找到”。
检查竞品。如果Personas提到替代方案，快速查看1-2个竞品网站，看他们是否更好地处理了相同问题。
外部曝光很重要。当Personas搜索公司时，会出现什么？评价？媒体报道？这些会在他们进入网站前就影响信任度。
焦点小组模式被低估。自由形式的反应往往能发现评分卡遗漏的点——情绪反应、意外的困惑、用户在调查中不会提及的问题。
一对一模式是核心用例。通过买家的视角将你的网站与竞品网站并列对比，能最快地明确你的定位差距。
此功能无代码脚本。通过Agent执行，使用WebSearch和WebFetch工具。

Dependencies

依赖项

Pre-built personas (
```
icp-persona-builder
```
output)
Web search capability (for external presence checks, competitor research)
Web fetch capability (for reading website pages)
No API keys or paid tools required

预构建的Personas (
```
icp-persona-builder
```
输出结果)
网页搜索能力（用于外部曝光检查、竞品研究）
网页抓取能力（用于读取网站页面）
无需API密钥或付费工具