For Agents · MCP

Your task.
The right skill.

The tunnel uses the same cards as the catalogue. Browse only as deep as needed — or load a broad bundle immediately.

No local installationNo M11 loginRead-only

Broad task

Load 5–10 and go

SEO, Sales, Agents or another broad area → one bundle call → work.

MCP endpoint: https://skills.m11.ch/mcp

Read-only access to published skills. Default 8, maximum 10 skills / 120,000 characters.

Task Packs

Marketing Foundation

Compact two-skill starter: clarify positioning and choose a lead magnet. Use Marketing Launch for the broader eight-skill go-to-market workflow.

Marketing Launch

Build an evidence-led marketing plan from ICP and competition through positioning, campaigns, growth and measurement.

Agent Audit Essentials

Diagnose architecture and context, plan agent-team responsibilities, then organize project context and session handoffs. Memory and cost-runtime reviews remain outside this pack.

Conversion and Activation Review

Review the journey from landing page and lead capture through registration, first value and transparent upgrades.

Campaign Content and Brand Review

Plan a campaign, draft its channel content and review the work against actual brand guidance.

Content and Launch Planning

Prioritize an editorial roadmap and plan how to launch and distribute it across suitable channels.

Customer, Market and Community Research

Understand customer needs, compare competitors and plan a community around real member value.

Lead Magnet to Lifecycle Nurture

Choose a relevant lead magnet, then draft a permission-based nurture journey with entry, suppression and exit rules.

API Contract and Observability

Define the API contract, then plan how to observe its latency, failures and retries. Guidance and checklist; no production changes.

Data Analysis Foundation

Profile a dataset, choose and interpret statistical methods, then validate calculations and conclusions before sharing.

Choose an area

01 · Category

What the MCP returns at this step

68 · Marketing · Ads Analysis

Meta Account Performance Analysis

Analyze Meta account performance, baselines and active-entity metrics. Use Meta Ads Deep Analysis for the wider audit framework.

metakontoleistungmeta-performance-analysis

Meta Ads Performance Analysis · Original SKILL.md

---
name: meta-performance-analysis
description: "Use when analyzing Meta (Facebook/Instagram) ad performance: account type detection, Pareto analysis, baselines, performance thresholds, issue diagnosis."
---
# Meta Ads - Performance Analysis Workflow

> Metric definitions (CTR, CPM, ROAS, CPL, CPR, conversion rate, account-type → PCM mapping, performance benchmarks, Pareto definition) live in `meta/tool-fundamentals` and auto-load with this skill. Do NOT redefine them here. This skill is the workflow for performance audits after the metrics are loaded.

## Step 1: Identify Account Type & PCM

Read `promoted_object.custom_event_type` from the ad set to determine the Primary Conversion Metric (PCM). The mapping table is in `tool-fundamentals → Account Type → Primary Conversion Metric`.

Where to read it:
- `facebook_get_details_of_ad_account` → returns `account_structure.adsets` with `promoted_object` for top spending ad sets. Use this first.
- `facebook_get_adset_details` → use when you need conversion events for ad sets not in the account structure.

***PROHIBITED***: Never extract or infer conversion events from names. Ad set names frequently don't match the actual config.

**Mixed accounts**: group ad sets by their `custom_event_type`. Each group gets its own PCM. Do not mix conversion types.

**If `promoted_object` is missing**: ASK the user which conversion event is the business outcome.

## Step 2: Pareto Pull (90% Spend)

1. Get ad-level insights sorted by spend descending.
2. Calculate cumulative spend percentage for each ad.
3. Keep ads where cumulative % ≤ 90% — these are your Pareto ads.
4. Focus analysis on Pareto ads.

**Query for Pareto ads** (MUST include active status filter):
```
Tool: facebook_get_adaccount_insights
level: "ad"
fields: ["ad_name", "ad_id", "adset_id", "adset_name", "spend", "impressions", "cpm", "ctr", "clicks", "actions", "action_values", "purchase_roas"]
date_preset: "last_30d"
filtering: [
  {"field": "impressions", "operator": "GREATER_THAN", "value": 0},
  {"field": "ad.effective_status", "operator": "IN", "value": ["ACTIVE"]}
]
sort: "spend_descending"
```

**CRITICAL**: Always include `ad.effective_status` (or `adset.effective_status`) filter. Without it, the API returns data for paused/deleted entities that had historical spend, producing analysis that includes inactive entities. Fetch ALL pages before analyzing — ad-level results paginate at 25 per page (see `tool-fundamentals → Pagination`).

## Step 3: Establish Baselines

From the Pareto set, compute:
- `avg_cpm` = mean CPM across Pareto ads
- `avg_ctr` = mean CTR across Pareto ads
- `avg_pcm` = mean ROAS (e-commerce) or mean CPL/CPR (lead gen / custom) across Pareto ads

These become the comparison points for individual-ad assessment. Apply the variance benchmarks from `tool-fundamentals → Performance Benchmarks`.

## Step 4: Diagnose Issues

**High CPM + Low PCM**:
- Expensive audience, not converting
- Action: pause ad or test different targeting

**Low CTR + Low PCM**:
- Ad not resonating with audience
- Check comments for negative sentiment
- If comments fine: test copy/creative variations with stronger CTA

**Good metrics but declining trend**:
- Check frequency — see fatigue signal in `tool-fundamentals → Diagnostic Signals`
- Action: new creative variations, not just budget changes

## Step 5: Generate Recommendations

**MANDATORY**: After completing the analysis, generate specific, actionable recommendations for each Pareto ad and for the account overall.

### Per-Ad Recommendations
For each Pareto ad, based on its diagnostic profile:
- **What to do**: specific action (pause, scale, create variation, change targeting)
- **Why**: metrics-based justification (e.g., "CPM 40% above Pareto avg with 0.3× ROAS")
- **How**: concrete next steps

### Account-Level Recommendations
Summarize the top 3–5 strategic recommendations:
1. Budget reallocation (shift spend from poor → good performers; respect guardrails)
2. Creative strategy (what's working, what to test next)
3. Targeting adjustments (if CPM issues are widespread)
4. Structure changes (campaign / ad set consolidation if needed)

### Output Format

```markdown
## Performance Summary

| Ad | Format | Spend | CPM | CTR | ROAS/CPA | Status |
|----|--------|-------|-----|-----|----------|--------|
| [Name] | Image | $X | $X | X% | X | [Scale/Pause/Test] |

**Benchmarks (Pareto avg)**: CPM: $X | CTR: X% | ROAS: X

## Recommendations

### Per-Ad Actions
1. **[Ad Name]**: [Action] — [Justification]
2. **[Ad Name]**: [Action] — [Justification]

### Strategic Recommendations
1. [Actionable recommendation with metrics basis]
2. [Actionable recommendation with metrics basis]
3. [Actionable recommendation with metrics basis]
```

**If creative analysis is needed** (video performance, hook rates, ad copy evaluation):
- The Creative Analysis skill should already be loaded for audits.
- If not loaded, use the `meta-creative-analysis` skill.

When to use

Analyze Meta account performance, baselines and active-entity metrics. Use Meta Ads Deep Analysis for the wider audit framework.

What you get

The complete original GoMarble guidance, with source attribution and declared limits.

How it works

Clarify scope, use available evidence, and apply the relevant source workflow within actual user authorization.

Requirements

Account type, conversion definitions, reporting period, targets and complete permitted report pages.

Delivery and review notes

An active-only filter answers current-active performance, not the full historical account; label selection and pagination coverage. GoMarble connectors, referenced sibling skills and runtime dependencies are not bundled or installed by M11. Brief obvious-danger screening only; no functional test or comprehensive safety certification. Loading this text authorizes no external calls, account mutations or ad spend.

Starting prompt

Use Meta Account Performance Analysis for [TASK]. Clarify Account type, conversion definitions, reporting period, targets and complete permitted report pages. Separate observed evidence from heuristics and recommendations. Check actual connector availability and approval semantics before any external action.

Not for

Claiming installed GoMarble tools, proven campaign outcomes or authorization to launch ads merely from loading this source.

What M11 added

German routing and explicit scope. Original attribution: GoMarble. An active-only filter answers current-active performance, not the full historical account; label selection and pagination coverage. GoMarble connectors, referenced sibling skills and runtime dependencies are not bundled or installed by M11. Brief obvious-danger screening only; no functional test or comprehensive safety certification. Loading this text authorizes no external calls, account mutations or ad spend.

Original authorship remains with gomarble-ai/marketing-agent · Original source ↗

Original license & copyright

MIT License

Copyright (c) 2026 GoMarble

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Source SHA-256: 3e1e11114408b3c81737efdf3cab8f08304962e1dee92992455f0dcbcc274d4a
Snapshot checked: 2026-09-28T16:19:52.867Z

Find a skill.Get to work.

All skills

Multi-Agent Orchestration

AI Context Window Audit

AI Agent Architecture Audit

AI Skill Discovery

Lead Magnet Strategy

Ideal Customer Profile

Go-to-Market Strategy

Marketing Campaign Ideas

Product Growth Loops

Competitor Analysis

North Star Metric

Product Positioning

Product Vision

Go-to-Market Motion Selection

Customer Feedback and JTBD Analysis

PESTLE Market Environment Analysis

Customer Journey Mapping

Ansoff Growth Options

Data Analysis Validation

Dataset Profiling

Statistical Analysis Guidance

Programmatic SEO Planning

Landing Page and Form Conversion Review

Paywall and Upgrade Planning

Signup and Registration Review

User Onboarding and Activation

Popup and Modal Planning

Email Sequence Copy and Flow

Marketing Campaign Planning

Marketing Content Drafting

Brand Voice and Content Review

Marketing Performance Reporting

Sales Company Research

Company and Contact Enrichment

Website Information Architecture

Content Strategy and Editorial Roadmap

Product Launch Planning

Customer Research and Voice of Customer

Community Growth and Member Experience

Competitor Research Profiles

Roadmap and Release Communication

Lifecycle Email Sequences

API Contracts and Interface Design

Observability and Instrumentation Planning

Agent Context and Session Handoff

Product Discovery Sprint

Deal Quality Scoring

Enrichment Waterfall Design

Customer Retention Playbook

Sales Coaching Practice

Revenue Cohort Analysis

Intent Signal Scoring

Customer Identity Matching

Segment Activation Planning

Sales Call Review

Retention Dashboard Design

Sales Coaching Competencies

Offer Delivery Design

Offer Value Communication

Service Productization

Conversion Hypothesis Prioritization

Braze Engagement Guidance

Brand Voice Guide Design

Content Pipeline Orchestration

Website Brand Profile

Buyer Objection Clarification

Klaviyo Integration Guidance

Offer Bonus Planning

Klaviyo Marketing Review

Cross-Platform Advertising Review

Expertise Business Model

LinkedIn Advertising Guidance

TikTok Advertising Guidance

Evidence Research Brief

Cinematic Video Prompts

Visual Prompt Model Adaptation

Visual Prompt Diagnostics

Short Film Development

Find a skill.
Get to work.