meta-ads-audit

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Meta Ads Audit

Meta Ads 审计

Diagnose Meta (Facebook + Instagram) account health and persist business context for downstream skills (

/meta-ads

). Read-only — never mutates the account. The user runs

/meta-ads

to execute fixes you recommend.

诊断Meta（Facebook + Instagram）账户健康状况，并为下游技能（

/meta-ads

）留存业务上下文。只读模式——绝不会修改账户内容。用户需运行

/meta-ads

来执行你推荐的修复操作。

Setup

设置步骤

../shared/preamble.md

— MCP detection, OAuth, ad account selection.

遵循

../shared/preamble.md

中的说明——MCP检测、OAuth授权、广告账户选择。

Filesystem contract (MUST persist)

文件系统约定（必须持久化）

Artifact	Path	When
Business context	`{data_dir}/meta/business-context.json`	First full audit, or refresh when `audit_date` is >90 days old. Skip on scoped audits if file is fresh.
Personas	`{data_dir}/meta/personas/{accountId}.json`	Every full audit.

These are the handoff to

/meta-ads

— write them even if the report itself is short. Otherwise downstream skills operate without business context and produce generic output.

If a

{data_dir}/business-context.json

exists from

/google-ads-audit

(no

meta/

subdir), read it as a starting point — most fields (services, brand voice, differentiators, locations, seasonality) are platform-agnostic. Then write the Meta-specific version to

{data_dir}/meta/business-context.json

with any Meta-specific overrides (different creative angles, different audiences, different funnel events).

business-context.json schema (shared with Google Ads where fields apply):

business_name, industry, website, services[], locations[], target_audience, brand_voice{tone, words_to_use[], words_to_avoid[]}, differentiators[], competitors[], seasonality{peak_months[], slow_months[], seasonal_hooks[]}, social_proof[], offers_or_promotions[], landing_pages{}, unit_economics{aov_usd, profit_margin, ltv_usd, source}, notes, audit_date, account_id

Meta-specific extensions:

meta_funnel_events{top_of_funnel, mid_of_funnel, conversion}, creative_inventory{concepts[], formats[], aspect_ratios[]}, custom_audiences{purchasers, abandoners, engagers, list_uploads[]}, pixel_health{pixel_id, capi_enabled, emq_score, last_event_at}

personas JSON schema:

{account_id, saved_at, personas: [{name, demographics, primary_goal, pain_points[], decision_trigger, value, meta_creative_angles[], visual_cues[]}]}

. The Meta version adds

meta_creative_angles

(e.g. "before/after demonstration", "founder-led explainer", "UGC review") and

visual_cues

(objects, settings, emotions that resonate with this persona). See

references/persona-discovery.md

产物	路径	时机
业务上下文	`{data_dir}/meta/business-context.json`	首次完整审计时，或当 `audit_date` 超过90天时刷新。若文件为最新状态，范围限定型审计可跳过此步骤。
用户画像	`{data_dir}/meta/personas/{accountId}.json`	每次完整审计时生成。

这些文件是传递给

/meta-ads

的关键信息——即使报告本身篇幅较短，也必须写入这些文件。否则下游技能将缺乏业务上下文，只能生成通用输出。

如果存在来自

/google-ads-audit

的

{data_dir}/business-context.json

文件（无

meta/

子目录），可将其作为起始参考——大多数字段（服务、品牌调性、差异化优势、地域、季节性）是平台无关的。随后将Meta特定版本写入

{data_dir}/meta/business-context.json

，并添加Meta专属的覆盖配置（如不同的创意方向、受众群体、漏斗事件）。

business-context.json schema（与Google Ads共享适用字段）：

business_name, industry, website, services[], locations[], target_audience, brand_voice{tone, words_to_use[], words_to_avoid[]}, differentiators[], competitors[], seasonality{peak_months[], slow_months[], seasonal_hooks[]}, social_proof[], offers_or_promotions[], landing_pages{}, unit_economics{aov_usd, profit_margin, ltv_usd, source}, notes, audit_date, account_id

。

Meta专属扩展字段：

meta_funnel_events{top_of_funnel, mid_of_funnel, conversion}, creative_inventory{concepts[], formats[], aspect_ratios[]}, custom_audiences{purchasers, abandoners, engagers, list_uploads[]}, pixel_health{pixel_id, capi_enabled, emq_score, last_event_at}

。

用户画像JSON schema：

{account_id, saved_at, personas: [{name, demographics, primary_goal, pain_points[], decision_trigger, value, meta_creative_angles[], visual_cues[]}]}

。Meta版本新增了

meta_creative_angles

（例如“前后对比展示”、“创始人讲解”、“UGC测评”）和

visual_cues

（与该用户画像共鸣的物品、场景、情绪）。详情请参阅

references/persona-discovery.md

。

Policy freshness check (run first)

政策新鲜度检查（优先执行）

Read

../shared/policy-registry.json

. For each entry where

last_verified + stale_after_days < today

High-volatility → WebSearch the
```
area
```
for recent Meta Ads changes; compare to
```
assumption
```
. If drift, banner the report and suggest registry update.
Moderate-volatility → one-line "may warrant a check" note.
Stable → skip silently.

The Meta platform changes faster than Google Ads (Advantage+, attribution, learning behaviors) — check high-volatility entries every audit.

读取

../shared/policy-registry.json

文件。对于每个满足

last_verified + stale_after_days < today

的条目：

高波动性 → 针对
```
area
```
进行Web搜索，查找Meta Ads的最新变化；与
```
assumption
```
进行对比。若存在偏差，在报告中标记并建议更新注册表。
中等波动性 → 添加一行“可能需要检查”的注释。
稳定性 → 静默跳过。

Meta平台的变化速度快于Google Ads（如Advantage+、归因模型、学习行为）——每次审计都需检查高波动性条目。

Phase 1 — Pull the audit dataset

阶段1 — 提取审计数据集

Use a single

runScript

call with

ads.graphParallel

to fan out the queries an audit needs. Build the fan-out from this rubric.

A complete audit needs at minimum:

Ad account info (
```
/{accountId}
```
) — currency, timezone, business id, spend cap, account status, balance.
Pixel health (
```
/{accountId}/customconversions
```
+
```
/{accountId}/adspixels
```
) — pixel id, last activity, CAPI status, Event Match Quality (EMQ) score.
Campaigns (
```
/{accountId}/campaigns
```
) — id, name, objective, status, daily/lifetime budget, special_ad_categories, buying_type, bid_strategy, created_time. Last 90 days.
Ad sets (
```
/{accountId}/adsets
```
) — id, name, status, campaign_id, optimization_goal, billing_event, bid_strategy, daily_budget, lifetime_budget, attribution_spec, targeting (summary), promoted_object, learning_stage_info.
Ads (
```
/{accountId}/ads
```
) — id, name, status, ad set, creative summary (image/video, primary text, headline, description, CTA), effective_status.
Insights at campaign level (
```
ads.insights({level:"campaign", date_preset:"last_30d"})
```
) — spend, impressions, reach, frequency, cpm, link CTR, link clicks, purchases (or other primary action), purchase value, ROAS, CPA.
Insights at ad set level — same fields, last 30 days.
Insights at ad level — top 50 ads by spend; same fields plus video metrics (3-sec views, ThruPlays) for video creatives.
Insights with breakdowns — placement (
```
publisher_platform,platform_position
```
), age/gender, device. Use these to spot placement losers and audience composition.
Recent edit activity — when available via
```
/{adsetId}
```
last_modified or
```
/{adsetId}
```
change history.

Compute aggregates in the script, return summarized JSON. Don't return all rows — rank, slice, summarize. The agent narrates the result; the script does the math.

suggestImprovement

is a useful cross-check for the server's heuristic surface — call it as a separate tool after the runScript pass if you want to compare your findings.

If a critical query errors out (auth, schema, API version), surface the error and stop — don't fall back to a degraded audit.

Skip scoring entirely if

totalSpend == 0

activeCampaigns == 0

. Go straight to business context.

使用单个

runScript

调用搭配

ads.graphParallel

来并行执行审计所需的查询。按照以下规则构建并行任务。

完整审计至少需要以下数据：

广告账户信息 (
```
/{accountId}
```
) — 货币、时区、企业ID、支出上限、账户状态、余额。
像素健康度 (
```
/{accountId}/customconversions
```
+
```
/{accountId}/adspixels
```
) — 像素ID、最后活动时间、CAPI状态、事件匹配质量（EMQ）分数。
广告系列 (
```
/{accountId}/campaigns
```
) — ID、名称、目标、状态、日预算/终身预算、特殊广告类别、购买类型、出价策略、创建时间。取最近90天的数据。
广告组 (
```
/{accountId}/adsets
```
) — ID、名称、状态、广告系列ID、优化目标、计费事件、出价策略、日预算、终身预算、归因规则、定向（摘要）、推广对象、学习阶段信息。
广告素材 (
```
/{accountId}/ads
```
) — ID、名称、状态、所属广告组、创意摘要（图片/视频、主文本、标题、描述、CTA）、有效状态。
广告系列层级洞察数据 (
```
ads.insights({level:"campaign", date_preset:"last_30d"})
```
) — 支出、曝光量、覆盖人数、频次、CPM、链接点击率、链接点击量、购买量（或其他主要转化动作）、购买价值、ROAS、CPA。
广告组层级洞察数据 — 相同字段，取最近30天的数据。
广告素材层级洞察数据 — 按支出排名前50的广告素材；相同字段，视频创意额外包含视频指标（3秒观看量、ThruPlays）。
带细分维度的洞察数据 — 投放位置（
```
publisher_platform,platform_position
```
）、年龄/性别、设备。用于识别表现不佳的投放位置和受众构成。
最近编辑活动 — 若
```
/{adsetId}
```
的last_modified字段或
```
/{adsetId}
```
的变更历史可用，则提取该数据。

在脚本中计算聚合数据**，返回汇总后的JSON。不要返回所有原始行——需进行排名、截取、汇总。由Agent解读结果，脚本负责计算。

suggestImprovement

是一个有用的交叉检查工具——如果需要对比你的发现，可在runScript执行完成后单独调用该工具。

若关键查询出错（授权、schema、API版本问题），需显示错误并停止审计——不要降级执行审计。

若满足以下条件则完全跳过评分：

totalSpend == 0

或

activeCampaigns == 0

。直接进入业务上下文阶段。

Phase 2 — Scope handling

阶段2 — 范围处理

If the user narrows the audit ("focus on one campaign", "campaign X", "just check creative fatigue"):

Match campaign names by case-insensitive substring. If no match, list available campaigns and ask.
Filter the in-memory dataset before scoring — no extra API calls.
Account-level dimensions (Pixel health, attribution defaults) stay account-wide. Note "Scoped to: X" in the report.
Skip Phase 4 (business context refresh) on scoped audits if
```
business-context.json
```
is fresh.

若用户限定审计范围（如“聚焦某一个广告系列”、“广告系列X”、“仅检查创意疲劳”）：

通过不区分大小写的子字符串匹配广告系列名称。若无匹配项，列出可用的广告系列并询问用户。
在评分前过滤内存中的数据集——无需额外调用API。
账户级维度（像素健康度、归因默认设置）仍保持账户范围。在报告中注明“审计范围：X”。
若
```
business-context.json
```
为最新状态，范围限定型审计可跳过阶段4（业务上下文刷新）。

Phase 3 — Score

阶段3 — 评分

Score each of the 7 dimensions 0–5 using

references/account-health-scoring.md

. Overall =

round(sum × 100 / 35)

Score	Label	Meaning
0	Critical	Broken or missing — actively losing money
1	Poor	Major waste or missed opportunity
2	Needs Work	Several clear issues
3	Acceptable	Functional, room to improve
4	Good	Well-managed, minor opportunities
5	Excellent	Best-practice

Scope-aware: campaign-level dimensions reflect in-scope data; account-level dimensions (Pixel + CAPI, attribution setup) score account-wide with a note on scope impact.

使用

references/account-health-scoring.md

中的规则，对7个维度分别进行0–5分的评分。总分 =

round(sum × 100 / 35)

。

分数	标签	含义
0	严重问题	功能损坏或缺失——正在持续亏损
1	较差	存在重大浪费或错失关键机会
2	需要改进	存在多个明确问题
3	可接受	功能正常，但有优化空间
4	良好	管理完善，仅存在少量优化机会
5	优秀	符合最佳实践

评分需考虑审计范围：广告系列级维度反映限定范围内的数据；账户级维度（像素+CAPI、归因设置）按全账户评分，并注明范围对评分的影响。

Encoded heuristics — apply these, they aren't obvious

内置启发式规则——需严格应用，这些规则并不直观

Pixel + CAPI is upstream of everything. EMQ < 7.0 means Meta can't match events well — Smart Bidding starves regardless of how good the creative is. STOP-condition input.
Reported ROAS systematically overstates true ROAS. Cross-check Meta-reported numbers against Shopify / GA4 / MMM where possible. The gap is the modeled-conversion premium and is typically 20–40% in ecom.
Frequency × CPM trend = creative diagnosis. Frequency > 3.0 with CPM rising ≥ 30% w/w is fatigue — recommend creative refresh, not budget cuts.
One ad set carrying > 70% of a campaign is fragility, not concentration. When it fatigues, the campaign collapses.
Audience overlap > 50% between sibling ad sets fragments signal. Consolidate; don't try to "fix" with bid caps.
Special Ad Category misclassification is a takedown risk, not just a policy nit. Surface as Critical regardless of current performance.
Manual placements without evidence is a sign of inherited-from-2018 thinking. Default should be Advantage+ Placements; deviations need data.

像素+CAPI是所有环节的基础。EMQ < 7.0意味着Meta无法准确匹配事件——无论创意质量多好，智能出价都会受限。此为停止条件输入项。
报告的ROAS会系统性高估真实ROAS。尽可能将Meta报告的数据与Shopify / GA4 / MMM的数据交叉验证。两者的差距即为模型转化溢价，在电商领域通常为20–40%。
频次×CPM趋势 = 创意诊断。频次>3.0且CPM周环比上升≥30%表示创意疲劳——建议刷新创意，而非削减预算。
单个广告组贡献某广告系列超过70%的支出是脆弱性表现，而非集中化优势。当该广告组出现疲劳时，整个广告系列会崩溃。
兄弟广告组间受众重叠>50%会分散信号。应合并广告组；不要尝试用出价上限“修复”该问题。
特殊广告类别分类错误存在被下架风险，而非仅仅是政策细节问题。无论当前表现如何，均标记为严重问题。
无数据支撑的手动投放位置选择是过时的操作思路。默认应使用Advantage+ Placements；若需调整，必须有数据支持。

Pixel + Tracking Diagnosis Matrix

像素与追踪诊断矩阵

	EMQ < 5	EMQ 5–6.9	EMQ 7.0+
CAPI off	Critical — flying blind	Critical — most events lost	High — leaving 15–25% of events on the table
CAPI on, dedup off	Critical — duplicated and weak signal	High — duplicate counting risk	Medium — match quality improves with dedup
CAPI on, dedup on	High — match quality is the bottleneck	Medium — improve event_id coverage	Healthy

	EMQ < 5	EMQ 5–6.9	EMQ 7.0+
CAPI未开启	严重问题——盲目投放	严重问题——大部分事件丢失	高优先级——损失15–25%的事件数据
CAPI已开启，去重未开启	严重问题——数据重复且信号薄弱	高优先级——存在重复计数风险	中等优先级——开启去重可提升匹配质量
CAPI已开启，去重已开启	高优先级——匹配质量是瓶颈	中等优先级——提升event_id覆盖率	健康状态

Phase 4 — Business context

阶段4 — 业务上下文

Derive what you can from the data already pulled:

Field	Source
`business_name`	Ad account name ( `/{accountId}` `name` field)
`services`	Top campaigns by spend, ad set names, top-converting ad creatives
`locations`	Targeting geo summary (countries / regions in active ad sets)
`brand_voice`	Top-performing ad copy (primary text + headline)
`creative_inventory.formats`	Mix of image / video / carousel observed in active ads
`creative_inventory.aspect_ratios`	Aspect ratios across active ads (1:1, 4:5, 9:16)
`meta_funnel_events.conversion`	Most common optimization event on top-spending ad sets
`custom_audiences`	Custom audiences referenced in active ad set targeting
`pixel_health`	From the Pixel detail call
`website`	Apex domain from active ad final URLs

Then crawl the website (homepage + about + 1–2 top landing pages, parallel

WebFetch

) and merge into the schema. See

references/business-context.md

for the full crawl procedure.

Always ask the user: differentiators, competitors, seasonality, AOV + profit margin (essential for ROAS-aware scoring). Ask for everything else only if data + crawl can't answer it.

从已提取的数据中推导尽可能多的信息：

字段	来源
`business_name`	广告账户名称（ `/{accountId}` 的 `name` 字段）
`services`	支出最高的广告系列、广告组名称、转化效果最好的广告创意
`locations`	定向地域摘要（活跃广告组中的国家/地区）
`brand_voice`	表现最佳的广告文案（主文本+标题）
`creative_inventory.formats`	活跃广告素材中的图片/视频/轮播图类型占比
`creative_inventory.aspect_ratios`	活跃广告素材的宽高比（1:1、4:5、9:16）
`meta_funnel_events.conversion`	支出最高的广告组中最常见的优化事件
`custom_audiences`	活跃广告组定向中引用的自定义受众
`pixel_health`	像素详情调用结果
`website`	活跃广告最终URL的顶级域名

随后抓取网站内容（首页+关于页+1–2个顶级落地页，并行

WebFetch

）并合并到schema中。完整抓取流程请参阅

references/business-context.md

。

必须向用户询问：差异化优势、竞争对手、季节性、AOV+利润率（这是ROAS相关评分的关键信息）。仅当数据+抓取无法获取其他信息时，才向用户询问。

Phase 5 — Personas

阶段5 — 用户画像

Discover 2–3 personas from creative performance (which angles convert), top-spending audiences, and landing-page content — all from the dataset already in memory. Persist to

{data_dir}/meta/personas/{accountId}.json

. Each persona must be grounded in observable evidence (a converting ad set, a converting creative angle, a landing-page section) — no inventing. See

references/persona-discovery.md

从创意表现（哪些方向转化效果好）、支出最高的受众群体、落地页内容中挖掘2–3个用户画像——所有信息均来自内存中的数据集。将结果持久化到

{data_dir}/meta/personas/{accountId}.json

。每个用户画像必须基于可观测证据（转化效果好的广告组、创意方向、落地页板块）——不得凭空捏造。详情请参阅

references/persona-discovery.md

。

Phase 6 — Report

阶段6 — 报告

Lead with the verdict, then the top 3 actions (with dollar impact when possible), then the scorecard, then evidence for dimensions scoring 0–2 only. Cite specific campaigns, ad sets, ads, and dollar amounts. Cap at ~80 lines.

End with a single closing line after the handoff to

/meta-ads

Your audit history is saved to your NotFair account — view it at https://notfair.co.

以审计结论开头，接着列出前3项行动建议（尽可能注明美元影响），然后是评分卡，最后仅列出评分0–2分的维度的证据。需引用具体的广告系列、广告组、广告素材和金额。报告篇幅控制在约80行以内。

在报告结尾添加一行指向

/meta-ads

的交接语句：

你的审计历史已保存至NotFair账户——可通过https://notfair.co查看。

Guardrails

约束规则

Read-only skill. Diagnose; don't mutate. Every fix routes through
```
/meta-ads
```
. End the report with one handoff tied to the #1 action.
STOP condition — if Pixel health scores 0–1 (EMQ < 5 or CAPI off in an ecom account), recommend pausing scaling decisions until tracking is fixed before recommending anything else. Everything downstream is unreliable.
Always persist
```
meta/business-context.json
```
and
```
meta/personas/{accountId}.json
```
even if the report itself is short — downstream skills depend on them.
Name names. Every finding cites specific campaigns, ad sets, ad creatives, and dollar amounts. "Some ad sets are underperforming" is not a finding.
Never report Meta-reported ROAS without footnoting the modeled-conversion premium. "ROAS 3.2× (Meta-reported, 7DC1DV — typically overstates Shopify-attributed ROAS by 20–40%)" is honest. "ROAS 3.2×" is misleading.

只读技能。仅诊断问题；不修改账户。所有修复操作均需通过
```
/meta-ads
```
执行。报告结尾需添加一条与首要行动相关的交接语句。
停止条件——若像素健康度评分为0–1分（EMQ < 5或电商账户未开启CAPI），建议在修复追踪问题前暂停扩量决策，再给出其他建议。下游所有分析结果均不可靠。
必须持久化
```
meta/business-context.json
```
和
```
meta/personas/{accountId}.json
```
，即使报告本身篇幅较短——下游技能依赖这些文件。
明确指向具体对象。每个发现都需引用具体的广告系列、广告组、广告素材和金额。“部分广告组表现不佳”不属于有效发现。
报告Meta的ROAS数据时必须注明模型转化溢价。例如“ROAS 3.2×（Meta报告，7天点击1天转化——通常比Shopify归因的ROAS高估20–40%）”是准确的表述。仅写“ROAS 3.2×”具有误导性。