google-ads-landing-review

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Google Ads Landing Review

Google Ads着陆页审查

Why this skill exists

此技能存在的原因

"The landing page isn't converting" is the most common complaint in Google Ads. But it hides two completely different root causes:

Tracking problem — conversions ARE happening but aren't being counted.
Path problem — the visitor arrives but the page fails them.

Most operators (and most clients) assume path failure when the numbers look bad. Half the time, it's a tracking failure. Conflating the two wastes months.

This skill separates them.

Read first:

google-ads/references/operator-thesis.md

google-ads/references/tracking-playbook.md

google-ads/references/structure-playbook.md

Read workspace if available:

```
workspace/ads/account.md
```
```
workspace/ads/goals.md
```
```
workspace/ads/findings.md
```
```
workspace/ads/drafts/_index.md
```
— check for existing tracking drafts

“着陆页无法转化”是Google Ads中最常见的投诉。但它隐藏了两种完全不同的根本原因：

追踪问题——转化确实发生了，但未被统计。
路径问题——访客到达页面，但页面未能满足其需求。

大多数运营人员（及大多数客户）在数据表现不佳时会默认是路径问题。但半数情况下，问题出在追踪上。混淆这两者会浪费数月时间。

此技能可将二者区分开来。

请先阅读：

google-ads/references/operator-thesis.md

google-ads/references/tracking-playbook.md

google-ads/references/structure-playbook.md

若工作区可用，请阅读：

```
workspace/ads/account.md
```
```
workspace/ads/goals.md
```
```
workspace/ads/findings.md
```
```
workspace/ads/drafts/_index.md
```
—— 检查是否存在现有追踪草稿

Diagnostic Model: Two Forks

诊断模型：两个分支

"Landing page isn't converting"
         │
    ┌────┴────┐
    │         │
  FORK A    FORK B
  Tracking  Path/UX
    │         │
  Is the     Is the page
  signal     actually
  correct?   failing?

Always run Fork A first. If tracking is broken, Fork B conclusions are unreliable.

"着陆页无法转化"
         │
    ┌────┴────┐
    │         │
  分支A    分支B
  追踪问题  路径/UX问题
    │         │
  数据信号   页面是否真的
  准确吗？   存在问题？

始终先运行分支A。若追踪存在问题，分支B的结论将不可靠。

Fork A: Tracking Diagnosis (Is the signal trustworthy?)

分支A：追踪诊断（数据信号是否可信？）

What to check

检查内容

1. Does a conversion action exist for the goal on this page?

Is there a conversion action configured for the form/call/purchase that this landing page is supposed to produce?
Is it set as primary (
```
include_in_conversions_metric = TRUE
```
)?
Is the counting type correct?

2. Does the tag actually fire?

Is the Google Ads conversion tag (or GA4 event imported to Google Ads) present on the confirmation/thank-you page?
Does the tag fire when the user completes the action? (Google Tag Assistant, network tab, or manual test)
Is the tag firing on the WRONG page? (e.g., firing on page load of the form page instead of the thank-you page)

3. Is the conversion path end-to-end intact?

Click on ad → landing page → form/CTA → thank-you page → tag fires → conversion recorded
Where does the chain break?

4. Cross-domain / redirect issues?

Does the landing page redirect to a different domain for the form/checkout?
If so, is cross-domain tracking configured?
Are UTM parameters / GCLID surviving the redirect?

5. Auto-tagging?

Is auto-tagging enabled?
Are there URL parameters being stripped by the landing page CMS or CDN?

6. Attribution window?

Conversion window too short (e.g., 1 day for a B2B lead that takes a week to decide)?
Multiple touchpoints lost?

1. 此页面的目标是否配置了转化动作？

针对该着陆页预期产生的表单提交/电话咨询/购买行为，是否配置了转化动作？
是否设置为主要转化动作（
```
include_in_conversions_metric = TRUE
```
）？
计数类型是否正确？

2. 标签是否真的触发？

确认页/感谢页上是否存在Google Ads转化标签（或导入至Google Ads的GA4事件）？
用户完成动作时标签是否触发？（可使用Google Tag Assistant、网络标签页或手动测试）
标签是否在错误页面触发？（例如，在表单页面加载时触发，而非感谢页）

3. 转化路径是否端到端完整？

点击广告 → 着陆页 → 表单/CTA → 感谢页 → 标签触发 → 转化被记录
链条在何处断裂？

4. 跨域/重定向问题？

着陆页是否会重定向至不同域名的表单/结账页面？
若是，是否配置了跨域追踪？
UTM参数/GCLID是否在重定向后保留？

5. 自动标记？

是否启用自动标记？
着陆页CMS或CDN是否会剥离URL参数？

6. 归因窗口？

转化窗口过短（例如，对于需要一周时间做决策的B2B线索，仅设置1天窗口）？
是否丢失了多个触点数据？

Fork A Data Acquisition (Connected Mode)

分支A数据获取（连接模式）

Conversion actions for this campaign:

sql

SELECT
  conversion_action.name,
  conversion_action.type,
  conversion_action.category,
  conversion_action.counting_type,
  conversion_action.include_in_conversions_metric,
  conversion_action.status,
  metrics.conversions,
  metrics.all_conversions
FROM conversion_action
WHERE segments.date DURING LAST_30_DAYS
  AND conversion_action.status = 'ENABLED'
ORDER BY metrics.conversions DESC

Campaign-level conversion data:

sql

SELECT
  campaign.name,
  campaign.final_url_suffix,
  metrics.clicks,
  metrics.conversions,
  metrics.all_conversions,
  metrics.cost_micros,
  metrics.cost_per_conversion
FROM campaign
WHERE campaign.status = 'ENABLED'
  AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.clicks DESC

Ad-level landing page URLs and performance:

sql

SELECT
  campaign.name,
  ad_group.name,
  ad_group_ad.ad.final_urls,
  ad_group_ad.ad.type,
  metrics.clicks,
  metrics.impressions,
  metrics.conversions,
  metrics.cost_micros
FROM ad_group_ad
WHERE campaign.status = 'ENABLED'
  AND ad_group_ad.status = 'ENABLED'
  AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.clicks DESC
LIMIT 50

Landing page experience (quality score indicators):

sql

SELECT
  campaign.name,
  ad_group.name,
  ad_group_criterion.keyword.text,
  ad_group_criterion.quality_info.quality_score,
  ad_group_criterion.quality_info.post_click_quality_score,
  ad_group_criterion.quality_info.creative_quality_score,
  ad_group_criterion.quality_info.search_predicted_ctr
FROM keyword_view
WHERE campaign.status = 'ENABLED'
  AND ad_group.status = 'ENABLED'
  AND ad_group_criterion.status = 'ENABLED'
ORDER BY ad_group_criterion.quality_info.post_click_quality_score ASC
LIMIT 50

此活动的转化动作：

sql

SELECT
  conversion_action.name,
  conversion_action.type,
  conversion_action.category,
  conversion_action.counting_type,
  conversion_action.include_in_conversions_metric,
  conversion_action.status,
  metrics.conversions,
  metrics.all_conversions
FROM conversion_action
WHERE segments.date DURING LAST_30_DAYS
  AND conversion_action.status = 'ENABLED'
ORDER BY metrics.conversions DESC

活动级转化数据：

sql

SELECT
  campaign.name,
  campaign.final_url_suffix,
  metrics.clicks,
  metrics.conversions,
  metrics.all_conversions,
  metrics.cost_micros,
  metrics.cost_per_conversion
FROM campaign
WHERE campaign.status = 'ENABLED'
  AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.clicks DESC

广告级着陆页URL及表现：

sql

SELECT
  campaign.name,
  ad_group.name,
  ad_group_ad.ad.final_urls,
  ad_group_ad.ad.type,
  metrics.clicks,
  metrics.impressions,
  metrics.conversions,
  metrics.cost_micros
FROM ad_group_ad
WHERE campaign.status = 'ENABLED'
  AND ad_group_ad.status = 'ENABLED'
  AND segments.date DURING LAST_30_DAYS
ORDER BY metrics.clicks DESC
LIMIT 50

着陆页体验（质量得分指标）：

sql

SELECT
  campaign.name,
  ad_group.name,
  ad_group_criterion.keyword.text,
  ad_group_criterion.quality_info.quality_score,
  ad_group_criterion.quality_info.post_click_quality_score,
  ad_group_criterion.quality_info.creative_quality_score,
  ad_group_criterion.quality_info.search_predicted_ctr
FROM keyword_view
WHERE campaign.status = 'ENABLED'
  AND ad_group.status = 'ENABLED'
  AND ad_group_criterion.status = 'ENABLED'
ORDER BY ad_group_criterion.quality_info.post_click_quality_score ASC
LIMIT 50

Fork A Verdict

分支A结论

Tracking Status	Meaning	Next Step
Clean	Tag fires correctly, conversion action configured right, GCLID passes	Proceed to Fork B
Suspicious	Tag exists but volume seems too low or too high vs. reality	Investigate specific break, then Fork B
Broken	No tag, wrong tag, or GCLID stripped	Fix tracking FIRST. Fork B is premature.
Unknown	Can't verify from API alone — needs manual tag inspection	Recommend Tag Assistant audit, then Fork B

追踪状态	含义	下一步
正常	标签触发正确，转化动作配置合理，GCLID正常传递	继续执行分支B
可疑	标签存在，但数据量与实际情况相比过高或过低	调查具体断裂点，再执行分支B
损坏	无标签、标签错误或GCLID被剥离	优先修复追踪问题。分支B暂不执行
未知	仅通过API无法验证——需要手动检查标签	建议使用Tag Assistant审计，再执行分支B

Fork B: Path/UX Diagnosis (Is the page actually failing?)

分支B：路径/UX诊断（页面是否真的存在问题？）

Only meaningful if Fork A shows tracking is Clean or Suspicious.

仅当分支A显示追踪状态为正常或可疑时，此诊断才有意义。

The Message Match Test

信息匹配测试

1. Search intent → Ad promise → Landing page delivery

The #1 conversion killer in Google Ads is message mismatch:

User searches "roll-off dumpster rental near me" (specific, purchase-intent)
Ad says "Container Solutions for Your Business" (vague)
Landing page is a generic homepage with 12 menu items

Each handoff is a potential drop:

Handoff	Question	Failure Mode
Search → Ad	Does the ad answer the specific search?	Generic ad for specific intent
Ad → Landing	Does the LP deliver what the ad promised?	"Request a Quote" ad → page with no form
Landing → CTA	Is the CTA visible, clear, and low-friction?	Form buried below fold, 15 fields
CTA → Completion	Can the user actually complete the action?	Broken form, redirect fails, captcha blocks

2. Specific checks (via browser or URL fetch):

Above-the-fold message: Does the H1/hero text match the ad's promise? Does it match the search intent?
CTA visibility: Can the user see what to do within 3 seconds? Is the CTA above the fold?
Form friction: How many fields? Required fields? Captcha? Multi-step?
Mobile experience: Does it work on mobile? (Most Google Ads clicks are mobile)
Page speed: Does it load in <3 seconds? (Slow = bounced)
Trust signals: Phone number, reviews, certifications, real photos?
Specificity: Does the page serve ONE intent, or is it a homepage trying to serve all intents?

1. 搜索意图 → 广告承诺 → 着陆页交付

Google Ads中转化率的头号杀手是信息不匹配：

用户搜索“附近的滚动式垃圾箱租赁”（具体、购买意图）
广告显示“为您的企业提供容器解决方案”（模糊）
着陆页是包含12个菜单项的通用首页

每个环节都可能导致用户流失：

环节	问题	失败模式
搜索 → 广告	广告是否回应了具体的搜索需求？	针对特定意图投放通用广告
广告 → 着陆页	着陆页是否兑现了广告的承诺？	广告显示“请求报价” → 页面无表单
着陆页 → CTA	CTA是否可见、清晰且低阻力？	表单隐藏在首屏下方，包含15个字段
CTA → 完成	用户能否真正完成动作？	表单损坏、重定向失败、验证码拦截

2. 具体检查（通过浏览器或URL抓取）：

首屏信息： H1/ hero文本是否与广告承诺匹配？是否符合搜索意图？
CTA可见性： 用户能否在3秒内看到需要执行的操作？CTA是否在首屏？
表单阻力： 有多少个字段？是否为必填字段？是否有验证码？是否为多步骤？
移动端体验： 在移动端能否正常使用？（大多数Google Ads点击来自移动端）
页面速度： 是否在3秒内加载完成？（加载慢会导致用户跳出）
信任信号： 是否有电话号码、评价、认证、真实照片？
针对性： 页面是否仅服务一种意图，还是试图满足所有意图的首页？

The Intent Routing Test

意图路由测试

3. Are different intent classes landing on different pages?

Intent Class	Should Land On	Common Failure
Buyer ("buy X now")	Product/service page with CTA	Homepage
Comparison ("X vs Y")	Comparison content	Product page with no comparison
Research ("how does X work")	Educational content	Sales page
Local ("X near me")	Location/service area page	National homepage
Brand ("company name")	Homepage or brand page	Generic product page

If multiple intent classes all route to the same generic page, that's a structure problem (→ recommend in structure draft), not a landing page problem.

3. 不同意图类别是否着陆到不同页面？

意图类别	应着陆页面	常见错误
购买者（“立即购买X”）	带有CTA的产品/服务页面	首页
对比者（“X vs Y”）	对比内容页面	无对比信息的产品页面
研究者（“X如何工作”）	教育内容页面	销售页面
本地用户（“附近的X”）	位置/服务区域页面	全国性首页
品牌用户（“公司名称”）	首页或品牌页面	通用产品页面

若多个意图类别均路由至同一通用页面，这属于结构问题（→ 建议在结构草稿中提出），而非着陆页问题。

The Conversion Path Walk

转化路径走查

4. Walk the actual path the user takes:

Click on ad
  → Landing page loads (check: speed, mobile rendering)
    → User reads headline (check: message match)
      → User finds CTA (check: visibility, clarity)
        → User clicks CTA (check: does it work?)
          → Form/checkout loads (check: friction, fields)
            → User submits (check: confirmation page loads)
              → Tag fires (check: conversion recorded)

Each step is a potential break. Document where the break is.

4. 走查用户实际经历的路径：

点击广告
  → 着陆页加载（检查：速度、移动端渲染）
    → 用户阅读标题（检查：信息匹配）
      → 用户找到CTA（检查：可见性、清晰度）
        → 用户点击CTA（检查：是否可用？）
          → 表单/结账页面加载（检查：阻力、字段）
            → 用户提交（检查：确认页面是否加载）
              → 标签触发（检查：转化是否被记录）

每个步骤都可能出现断裂。记录断裂位置。

Fork B Scoring

分支B评分

Rate each dimension:

Dimension	Score	Notes
Message match (search→ad→page)	Strong / Weak / Missing
CTA clarity	Clear / Buried / Missing
Form friction	Low (≤4 fields) / Medium (5-8) / High (9+)
Mobile experience	Good / Adequate / Broken
Page speed	Fast (<3s) / Slow (3-6s) / Broken (>6s)
Trust signals	Strong / Some / None
Intent specificity	Focused / Mixed / Generic
Conversion path completeness	Complete / Partially broken / Broken

对每个维度进行评分：

维度	评分	备注
信息匹配（搜索→广告→页面）	强 / 弱 / 缺失
CTA清晰度	清晰 / 隐藏 / 缺失
表单阻力	低（≤4个字段） / 中（5-8个） / 高（9+个）
移动端体验	良好 / 尚可 / 损坏
页面速度	快（<3秒） / 慢（3-6秒） / 损坏（>6秒）
信任信号	强 / 部分 / 无
意图针对性	聚焦 / 混合 / 通用
转化路径完整性	完整 / 部分断裂 / 断裂

Data Acquisition — Landing Page Review

数据获取 —— 着陆页审查

Connected Mode (MCP + Browser/Fetch)

连接模式（MCP + 浏览器/抓取）

Pull ad final URLs and campaign data via GAQL (see queries above)
For each unique landing URL:
- Fetch with
```
web_fetch
```
  for content analysis (H1, CTA text, form fields, page structure)
- Use
```
browser
```
  for interactive checks if needed (JavaScript-rendered pages, form testing)
- Check mobile rendering if concerns arise
Cross-reference landing pages against search term intent classes from
```
workspace/ads/intent-map.md
```

通过GAQL获取广告最终URL及活动数据（见上述查询）
针对每个唯一着陆URL：
- 使用
```
web_fetch
```
  抓取内容进行分析（H1、CTA文本、表单字段、页面结构）
- 若需要，使用
```
browser
```
  进行交互式检查（JavaScript渲染页面、表单测试）
- 若有疑问，检查移动端渲染情况
将着陆页与
```
workspace/ads/intent-map.md
```
中的搜索词意图类别进行交叉比对

Export Mode

导出模式

Ask the user for:

Landing page URL(s)
Which campaigns/ad groups point to which pages
Conversion action name and how it fires (page load, event, etc.)
Any known issues (form complaints, mobile problems)
Recent conversion volume (or "we don't know" — that's data too)

向用户索要：

着陆页URL
哪些活动/广告组指向哪些页面
转化动作名称及其触发方式（页面加载、事件等）
已知问题（表单投诉、移动端问题）
近期转化量（或“未知”——这也是数据）

Differential Diagnosis Summary

鉴别诊断总结

After running both forks, produce a clear classification:

运行两个分支后，生成明确的分类：

Scenario 1: Tracking Problem Masquerading as UX Problem

场景1：伪装成UX问题的追踪问题

Symptoms: Low/zero conversions, but page looks fine, form works, users seem to engage Diagnosis: Tag not firing, wrong conversion action, GCLID stripped, cross-domain break Action: Fix tracking → then reassess conversion rate with clean data Draft type: Tracking fix (use

drafts/templates/tracking-draft.md

)

症状： 转化率低/为零，但页面看起来正常，表单可用，用户似乎有参与度 诊断： 标签未触发、转化动作错误、GCLID被剥离、跨域断裂 行动： 修复追踪问题 → 然后使用干净的数据重新评估转化率 草稿类型： 追踪修复（使用

drafts/templates/tracking-draft.md

）

Scenario 2: UX/Path Problem (Tracking Is Fine)

场景2：UX/路径问题（追踪正常）

Symptoms: Tracking verified clean, but conversion rate is genuinely low Diagnosis: Message mismatch, buried CTA, excessive form friction, wrong page for intent Action: Landing page improvements or intent routing changes Draft type: Landing review draft (use template below) and/or structure draft

症状： 追踪已验证正常，但转化率确实偏低 诊断： 信息不匹配、CTA隐藏、表单阻力过大、页面与意图不匹配 行动： 优化着陆页或调整意图路由 草稿类型： 着陆页审查草稿（使用下方模板）和/或结构草稿

Scenario 3: Both Problems

场景3：两者皆有问题

Symptoms: Tracking has issues AND the page has UX problems Diagnosis: Two independent failures compounding Action: Fix tracking first (P0), then address UX (P1) — in that order Draft types: Tracking fix draft + Landing review draft

症状： 追踪存在问题且页面存在UX问题 诊断： 两个独立故障相互叠加 行动： 优先修复追踪问题（P0），然后解决UX问题（P1）——按此顺序 草稿类型： 追踪修复草稿 + 着陆页审查草稿

Scenario 4: Traffic Quality Problem (Page and Tracking Are Fine)

场景4：流量质量问题（页面与追踪均正常）

Symptoms: Tracking clean, page is good, but conversion rate still low Diagnosis: The traffic is wrong — keywords matching wrong intent, broad match pulling junk, PMax sending Display/YouTube traffic to a Search landing page Action: Search terms analysis, negative keywords, or structure changes Draft type: Negative draft and/or structure draft (this is NOT a landing page problem)

症状： 追踪正常，页面良好，但转化率仍偏低 诊断： 流量质量不佳——关键词匹配错误意图、广泛匹配引入无效流量、PMax将展示/YouTube流量导向搜索着陆页 行动： 搜索词分析、添加否定关键词或调整结构 草稿类型： 否定关键词草稿和/或结构草稿（这不属于着陆页问题）

Draft Output

草稿输出

Landing Review Draft

着陆页审查草稿

Trigger: Fork B finds meaningful path/UX issues (at least 2 dimensions scored Weak/Missing/Broken)

Create using the template below:

Write to

workspace/ads/drafts/YYYY-MM-DD-[account-slug]-landing-review.md

Update
```
workspace/ads/drafts/_index.md
```

If Fork A also finds issues, create a SEPARATE tracking fix draft — don't mix them.

触发条件： 分支B发现明显的路径/UX问题（至少2个维度评分为弱/缺失/断裂）

使用下方模板创建：

写入

workspace/ads/drafts/YYYY-MM-DD-[account-slug]-landing-review.md

更新
```
workspace/ads/drafts/_index.md
```

若分支A也发现问题，请创建单独的追踪修复草稿——不要将二者混合。

Landing Review Draft Template

着陆页审查草稿模板

markdown

undefined

markdown

undefined

Draft: Landing Page Review — [DATE]

草稿：着陆页审查 —— [日期]

Status: proposed Skill: /google-ads landing-review Account: [Customer ID / Name]

状态：提议中技能：/google-ads landing-review 账户：[客户ID / 名称]

Summary

摘要

[One paragraph: which pages were reviewed, the primary diagnosis (tracking vs path vs both), and the highest-priority fix.]

[一段文字：审查了哪些页面，主要诊断结果（追踪问题/路径问题/两者皆有/流量质量问题），以及最高优先级的修复建议。]

Diagnostic Classification

诊断分类

Primary issue: Tracking Problem | Path/UX Problem | Both | Traffic Quality Problem

主要问题： 追踪问题 | 路径/UX问题 | 两者皆有 | 流量质量问题

Fork A: Tracking Status

分支A：追踪状态

Conversion action: [Name and status]
Tag status: [Fires correctly / Suspicious / Broken / Unknown]
GCLID passing: [Yes / No / Unknown]
Auto-tagging: [Enabled / Disabled]
Verdict: [Clean / Suspicious / Broken / Unknown]

转化动作： [名称及状态]
标签状态： [触发正常 / 可疑 / 损坏 / 未知]
GCLID传递： 是 / 否 / 未知
自动标记： 启用 / 禁用
结论： 正常 / 可疑 / 损坏 / 未知

Fork B: Path/UX Assessment

分支B：路径/UX评估

Page: [URL]

页面：[URL]

Campaigns pointing here: [list]
Clicks (30d): [N]
Conversions (30d): [N]
Implied conversion rate: [X%]

指向此页面的活动： [列表]
点击量（30天）： [N]
转化量（30天）： [N]
隐含转化率： [X%]

Scores

评分

Dimension	Score	Detail
Message match	[Strong/Weak/Missing]	[specific observation]
CTA clarity	[Clear/Buried/Missing]	[specific observation]
Form friction	[Low/Medium/High]	[field count, issues]
Mobile experience	[Good/Adequate/Broken]	[specific observation]
Page speed	[Fast/Slow/Broken]	[load time if available]
Trust signals	[Strong/Some/None]	[specific observation]
Intent specificity	[Focused/Mixed/Generic]	[specific observation]
Path completeness	[Complete/Partial/Broken]	[where it breaks]

维度	评分	详情
信息匹配	[强/弱/缺失]	[具体观察结果]
CTA清晰度	[清晰/隐藏/缺失]	[具体观察结果]
表单阻力	[低/中/高]	[字段数量、问题]
移动端体验	[良好/尚可/损坏]	[具体观察结果]
页面速度	[快/慢/损坏]	[若有可用的加载时间]
信任信号	[强/部分/无]	[具体观察结果]
意图针对性	[聚焦/混合/通用]	[具体观察结果]
路径完整性	[完整/部分断裂/断裂]	[断裂位置]

Proposed Changes

建议修改

Change 1: [Specific recommendation]

修改1：[具体建议]

Current state: [what's wrong]
Proposed state: [what to do]
Expected impact: [on conversion rate, quality score]
Risk: [what could go wrong]
Priority: P0 / P1 / P2

[repeat for each change]

当前状态： [存在的问题]
提议状态： [解决方案]
预期影响： [对转化率、质量得分的影响]
风险： [可能出现的问题]
优先级： P0 / P1 / P2

[重复上述格式列出每个修改建议]

Intent Routing Assessment

意图路由评估

Are different intent classes landing on the right pages? [Yes / No — detail]
Should new landing pages be created? [If so, for which intent classes]
Cross-reference with intent map: [link to workspace/ads/intent-map.md findings]

不同意图类别是否着陆到正确页面？ 是 / 否 —— 详情
是否需要创建新的着陆页？若是，针对哪些意图类别
与意图映射交叉比对： [链接至workspace/ads/intent-map.md的发现结果]

Dependencies

依赖项

[e.g., "Fix tracking (2026-03-15-acme-tracking-fix.md) before assessing conversion rate improvement"]

[例如：“在评估转化率提升前，先修复追踪问题（2026-03-15-acme-tracking-fix.md）”]

Confidence

置信度

[High / Medium / Low] — [reasoning]

[高 / 中 / 低] —— [理由]

Review

审核

Evidence checked
Collateral risk checked
Dependencies checked
Decision: approve | defer | reject
Decision reason: ____
Reviewed by: ____
Reviewed on: ____
Applied on: ____
Notes: ____

undefined

证据已检查
相关风险已检查
依赖项已检查
决策： 批准 | 推迟 | 拒绝
决策理由： ____
审核人： ____
审核日期： ____
实施日期： ____
备注： ____

undefined

Always update workspace memory:

始终更新工作区记忆：

```
workspace/ads/findings.md
```
— landing page diagnosis and classification
```
workspace/ads/learnings.md
```
— what we learned about this account's conversion path
Update existing tracking drafts if Fork A reveals new tracking problems

```
workspace/ads/findings.md
```
—— 着陆页诊断及分类结果
```
workspace/ads/learnings.md
```
—— 关于此账户转化路径的经验总结
若分支A发现新的追踪问题，更新现有追踪草稿

Output Shape

输出结构

Account Status block — name, CID, mode, date range, tracking confidence
Diagnostic classification — tracking problem vs path problem vs both vs traffic quality
Fork A summary — tracking status for each page/campaign reviewed
Fork B summary — path/UX scores for each landing page reviewed
Differential diagnosis — which scenario applies (1, 2, 3, or 4)
Prioritized recommendations — fix order matters (always tracking before UX)
Drafts created — with file paths and summaries
Memory updates

账户状态块 —— 名称、CID、模式、日期范围、追踪置信度
诊断分类 —— 追踪问题/路径问题/两者皆有/流量质量问题
分支A摘要 —— 所审查的每个页面/活动的追踪状态
分支B摘要 —— 每个着陆页的路径/UX评分
鉴别诊断 —— 适用哪种场景（1、2、3或4）
优先级建议 —— 修复顺序很重要（始终先修复追踪问题，再处理UX）
已创建草稿 —— 文件路径及摘要
记忆更新

Rules

规则

Always run Fork A first. If tracking is broken, do NOT produce detailed UX recommendations — they will be based on phantom conversion data.
Distinguish clearly: A page with a broken tag and 0% conversion rate is a tracking problem, not a UX problem. Say it explicitly.
Don't blame the landing page for traffic quality problems. If the keywords are sending the wrong people, the page can be perfect and still not convert. Route to negatives/structure skills.
Walk the actual path. Don't just look at the page — follow the entire click → conversion chain.
Be specific. "The form has too many fields" is vague. "The form has 12 required fields including 'company size' and 'annual revenue' which are unnecessary for an initial quote request" is useful.
Mobile first. Most Google Ads clicks are mobile. If you can only check one thing, check mobile.
One page at a time. Don't try to review 10 pages at once. Start with the highest-spend page.
Message match is almost always the problem. When in doubt, check whether the H1 matches the search intent. If there's a mismatch, that's usually the answer.

始终先运行分支A。若追踪存在问题，请勿生成详细的UX建议——这些建议将基于虚假的转化数据。
明确区分： 标签损坏且转化率为0%的页面属于追踪问题，而非UX问题。请明确说明。
不要因流量质量问题指责着陆页。若关键词引入了错误的用户，即使页面完美也无法转化。请转至否定关键词/结构技能处理。
走查实际路径。不要仅查看页面——跟随整个点击→转化链条。
具体明确。“表单字段过多”过于模糊。“表单包含12个必填字段，包括‘公司规模’和‘年收入’，而这些对于初始报价请求来说并非必要”才有用。
移动端优先。大多数Google Ads点击来自移动端。若只能检查一项内容，请检查移动端体验。
一次审查一个页面。不要试图同时审查10个页面。从花费最高的页面开始。
信息匹配几乎总是问题所在。如有疑问，检查H1是否与搜索意图匹配。若存在不匹配，通常这就是问题的答案。