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

82 · Marketing · Ads Analysis

Google Shopping Optimization

Review Shopping feed health and product-level decisions before campaign changes; not PMax optimization.

shoppinganzeigenoptimierunggoogle-ads-shopping

Google Ads Shopping Optimization · Original SKILL.md

---
name: google-ads-shopping
description: "Use when analyzing or optimizing Google Ads Shopping campaigns: feed health gate, item KILL/DOWNGRADE/PROMOTE rules, product groups, search-term negatives, SCALE/T/CAP."
---
# Google Ads Shopping — Optimization Protocol

Product-led optimization: validate feed → classify items → manage structure → scale or cut.

## Step 0: Feed Health (BLOCKING — fail = STOP)

Before any optimization, verify:
- ≥ 80% products approved in Merchant Center
- Purchase conversion tracking is active (conversions_value > 0 in data)
- Price and availability are accurate (requires manual verification or Merchant Center check)

**If ANY check fails → STOP all optimization. Fix feed first.**

---

## Step 1: Data Pull

Query using `google_ads_run_gaql`:

1. **Account AOV**: total conversions_value / total conversions from `customer` resource, LAST_30_DAYS
2. **Item-level performance**: product_item_id, product_title, product_brand, cost, conversions, conv_value, clicks, impressions from `shopping_performance_view`. Sort by cost DESC, limit 200. Filter: `SHOPPING` campaigns.
3. **Campaign-level**: campaign id/name/status, budget, cost, conversions, conv_value from `campaign`. Filter: `SHOPPING`, `ENABLED`.
4. **Search terms**: search_term, cost, clicks, conversions from `search_term_view`. Filter: `SHOPPING`. Sort by cost DESC, limit 200.
5. **Daily budget utilization**: campaign id, date, cost from `campaign`. Filter: `SHOPPING`, `LAST_14_DAYS`.

**Lookback**: 14 days if ≥ 100 purchases/month, otherwise 30 days.

Ask user for **target ROAS**. Default: historical 30-day ROAS.

---

## Step 2: Item-Level Classification

Sort items by cost (highest first). For each item:

| Classification | Condition | Action |
|---------------|-----------|--------|
| **KILL** | Cost ≥ 2× AOV AND conversions = 0 | Exclude item ID immediately. No learning window. |
| **DOWNGRADE** | ROAS < 0.7× target AND cost ≥ 0.5× AOV | Lower bid or exclude. Item is spending but underperforming. |
| **PROMOTE** | Conversions ≥ 2 AND ROAS ≥ target | Isolate into own product group/campaign to protect budget. |

Items not matching any rule: monitor, no action needed.

---

## Step 3: Product Group Structure

- Maximum 3 product groups per campaign
- Split ONLY when economics differ (different margins, price bands, seasonal risk)
- If groups don't have meaningfully different ROAS targets → consolidate

---

## Step 4: Search Term Negatives

Add negatives ONLY for wrong-intent terms:
- Price objection: free, cheap, budget, discount code, coupon
- DIY/repair: diy, repair, fix, how to, tutorial
- Used/second-hand: used, second hand, refurbished, pre-owned
- Products not in feed (compare search terms to product titles)
- Irrelevant attributes (colors/sizes not offered)

Do NOT negate high-intent commercial terms even if they haven't converted yet.

---

## Step 5: Budget & Bid Decisions

### SCALE (increase budget 10–20%)
ALL must be true:
- Campaign ROAS ≥ target
- Budget utilization ≥ 90% on 10+ of last 14 days
- Top 20% of SKUs by conversions account for ≥ 60% of spend (winners getting the budget)

### CUT/CAP (reduce budget or exclude SKUs)
ANY true:
- Campaign ROAS < target
- SKUs with ROAS < 0.7× target receiving increasing spend share WoW

→ Reduce budget OR exclude underperforming SKUs first, then reassess.

---

## Step 6: Device & Location (last)

Only analyze if segment spend ≥ 2× AOV (enough data to judge).

- Exclude device/location ONLY if ROAS ≤ 0.6× account average AND no operational explanation (e.g., mobile site is broken)
- Check for technical issues before excluding (site speed, checkout flow)

---

## Step 7: Campaign Kill Conditions

Pause the Shopping campaign if:
- ROAS < 0.6× target after total spend ≥ 2× AOV
- Shopping ROAS consistently worse than Search ROAS over 30+ days

---

## Prohibitions

- Never optimize by CTR alone (Shopping is ROAS-driven)
- Never let cheap, low-converting SKUs dominate spend
- Never scale without confirming winners are getting the budget

Output must include: item classification summary (KILL/DOWNGRADE/PROMOTE counts + savings), search term negatives, budget recommendation with rationale, expected 7-day ROAS impact.

When to use

Review Shopping feed health and product-level decisions before campaign changes; not PMax optimization.

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

Shopping campaign data, product feed quality, margins, conversion evidence and targets.

Delivery and review notes

KILL/DOWNGRADE/PROMOTE are recommendations for review, not automatic actions; preserve margin and feed-quality evidence. 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 Google Shopping Optimization for [TASK]. Clarify Shopping campaign data, product feed quality, margins, conversion evidence and targets. 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. KILL/DOWNGRADE/PROMOTE are recommendations for review, not automatic actions; preserve margin and feed-quality evidence. 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: 28222cd7b34b8dd6b28a2209ecd68571ac4488a600a73c8569330091503047ea
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.