What is MCP Server for Google Ads?

MCP Server for Google Ads is a Model Context Protocol implementation that connects AI agents like Claude, Gemini, and ChatGPT directly to Google Ads API. It enables real-time campaign data access, performance analysis, and automated optimization through natural language queries instead of manual exports or complex API calls.

How do I install Google's official MCP Server for Google Ads?

Clone the repository from github.com/google-marketing-solutions/google_ads_mcp, install Python 3.12+ dependencies, configure OAuth2 credentials, and set up your Google Ads developer token. The official server provides list_accessible_customers and search tools for raw GAQL queries but requires technical knowledge of Google Ads Query Language.

What's the difference between official and managed MCP servers?

Google's official MCP server offers direct GAQL access but requires Python setup, OAuth management, and GAQL knowledge. Managed solutions like Ryze AI provide structured tools for campaign creation, keyword management, and reporting without technical setup. Official servers are free but complex; managed servers offer ease-of-use with potential costs.

Can MCP Server create and modify Google Ads campaigns?

Google's official MCP server is currently read-only for safety. Managed MCP solutions may offer write capabilities for campaign creation, keyword additions, and budget adjustments. Always verify permissions and start with read-only analysis before enabling automated campaign modifications that could impact ad spend.

Which AI agents work with Google Ads MCP Server?

Claude Desktop, Google Gemini, ChatGPT with plugins, and custom enterprise AI platforms support MCP integration. Claude Desktop offers the most seamless setup, Gemini has native Google ecosystem integration, and ChatGPT requires additional configuration. Any MCP-compatible AI agent can connect to Google Ads data.

Do I need a Google Ads developer token for MCP Server?

Yes, both official and managed MCP servers require Google Ads API access through a developer token. Apply through Google Ads Manager, complete the review process (typically 24-48 hours), and configure OAuth2 credentials. Managed solutions may handle token management but still need your account permissions.

GOOGLE ADS

MCP Server Google Ads — Connect Claude AI to Google Ads API (Complete 2026 Guide)

MCP server Google Ads integration connects Claude AI directly to Google Ads API through Model Context Protocol, enabling real-time campaign analysis, automated reporting, and optimization recommendations without manual data exports or GAQL syntax knowledge.

Ira Bodnar·April 30, 2026·Updated Apr 30, 2026·18 min read

Contents

Autonomous Marketing

Grow your business faster with AI agents

✓Automates Google, Meta + 5 more platforms
✓Handles your SEO end to end
✓Upgrades your website to convert better

What is MCP server Google Ads integration?

MCP server Google Ads integration is a standardized bridge that connects Large Language Models like Claude directly to your Google Ads account via the Model Context Protocol. Instead of manually exporting campaign data, creating reports in spreadsheets, and copying metrics between platforms, the MCP server enables Claude to query your advertising data in real-time, analyze performance trends, and generate optimization recommendations on demand.

Google released their official MCP server Google Ads implementation in March 2026, providing two core functions: list_accessible_customers and search. The search function accepts GAQL (Google Ads Query Language) queries, giving Claude direct access to any metric, dimension, or resource in your Google Ads account. Average setup time is 15-20 minutes for technical users familiar with Python environments and Google Cloud Console.

The critical difference from manual reporting: MCP server Google Ads connections provide live data, not stale exports. When Claude analyzes your > 50% impression share loss last week, it is pulling yesterday’s actual auction data — not a CSV you exported on Monday. This real-time access enables automated monitoring that catches budget depletion, bid cap issues, and quality score drops within hours, not days.

This guide covers the complete setup process, from Google Cloud authentication to Claude Desktop configuration, plus 8 automation workflows that eliminate 12-18 hours of weekly Google Ads management. For the broader context of using Claude with Google Ads, see How to Use Claude for Google Ads. For specific Claude skills beyond MCP integration, see Claude Skills for Google Ads.

1,000+ Marketers Use Ryze

Automating hundreds of agencies

★★★★★4.9/5

What are the 3 ways to set up MCP server Google Ads?

There are three approaches to connecting Claude with Google Ads data via MCP server, each with different tradeoffs between setup complexity, customization, and maintenance overhead. Google’s official MCP server requires GAQL knowledge and local Python setup, while managed solutions eliminate technical barriers at the cost of flexibility.

Method	Setup Time	Technical Skill	Best For
Google Official MCP Server	15-25 minutes	High (Python, GAQL)	Developers, custom queries
Ryze AI Managed MCP	Under 3 minutes	None (OAuth only)	Marketers, fast setup
Third-Party MCP Bundles	8-12 minutes	Medium (API keys)	Advanced users, multiple tools

Google Official MCP Server provides direct GAQL access through the googleads/google-ads-mcp repository. You clone the repo, install Python dependencies, configure OAuth credentials in Google Cloud Console, generate a developer token, and connect via Claude Desktop. Full control over queries but requires knowledge of Google Ads resource structure (campaigns table, ad_groups, keywords) and GAQL syntax.

Ryze AI Managed MCP eliminates setup complexity entirely. Sign up at get-ryze.ai/mcp, authenticate with Google Ads via OAuth, receive pre-configured MCP credentials, and paste them into Claude Desktop. No Python environment, no GAQL learning curve, no developer token application. Handles rate limiting and API updates automatically.

Third-Party MCP Bundles like MCPBundles offer structured functions beyond raw GAQL querying — campaign creation, RSA ad writing, keyword research, analytics reporting. Requires API configuration but provides English-language functions instead of SQL-like queries. Middle ground between Google’s technical approach and fully managed solutions.

Tools like Ryze AI automate this process — monitoring performance 24/7, adjusting bids in real-time, and optimizing budget allocation without manual intervention. Ryze AI clients see an average 3.2x improvement in ROAS within 8 weeks.

What are 8 Google Ads workflows you can automate with MCP server?

Google Ads MCP server enables seven core automation workflows that eliminate 10-15 hours of weekly manual tasks. These workflows cover campaign performance analysis, keyword research and optimization, bid strategy recommendations, budget allocation guidance, automated reporting, competitor analysis, and real-time optimization alerts. Each workflow leverages real-time API data to provide actionable insights within seconds.

Workflow 01

Search Term Performance Analysis

Search terms with < 1% conversion rate but > 100 clicks cost the average Google Ads account $2,000-4,000 monthly in wasted spend. Claude pulls search term reports via MCP, identifies high-cost zero-conversion queries, calculates waste by match type, and recommends negative keyword additions. It categorizes terms as immediate negatives (zero conversions, > $200 spend) or watch-list items (low performance but insufficient data).

Example promptAnalyze search terms for the last 30 days. Identify terms with >$100 spend and 0 conversions. Calculate wasted budget by match type. Recommend negative keywords to add at campaign and ad group levels.

Workflow 02

Quality Score Optimization

Quality Score directly impacts CPC — a drop from 8 to 6 can increase costs by 25-30%. Claude monitors keyword-level Quality Scores, identifies recent drops, correlates them with landing page experience and ad relevance metrics, and prioritizes keywords by potential CPC savings. It flags ad groups where > 50% of keywords have Quality Score < 6 and estimates cost impact.

Example promptPull Quality Score data for all keywords with >100 impressions last month. Flag keywords that dropped 2+ points. Group by reason (landing page, relevance, CTR). Estimate CPC impact and prioritize by potential savings.

Workflow 03

Automated Bid Cap Analysis

Bid caps set too low cause impression share loss; too high inflates CPCs without volume gains. Claude analyzes impression share lost due to bid caps versus rank, compares current max CPC to auction-time winning bids, and identifies keywords constrained by conservative bidding. It recommends bid adjustments to capture 10-15% more qualified traffic at target CPA.

Example promptCheck impression share lost to bid caps vs rank for all campaigns. Compare max CPC to top-of-page bid estimates. Identify keywords losing >30% impression share due to low bids. Recommend bid increases to capture more traffic.

Workflow 04

RSA Performance Monitoring

Responsive Search Ads with poor asset combinations see 20-40% higher CPA than optimized versions. Claude tracks RSA asset performance (headlines, descriptions), identifies low-performing combinations, flags assets with "Low" or "Don't show" ratings, and recommends new headlines/descriptions based on top performers. It also detects asset repetition and redundancy issues.

Example promptAudit RSA asset performance across all ad groups. Flag assets rated "Low" or "Don't show". Identify redundant headlines. Recommend 5 new headlines based on top- performing asset patterns.

Workflow 05

Weekly Performance Reporting

Manual Google Ads reporting takes 3-4 hours weekly: export data, create charts, write commentary, analyze trends. Claude generates executive-ready reports in under 90 seconds, including spend vs budget, CPA trends, conversion volume changes, top/bottom performing campaigns, and 3-5 specific optimization recommendations with estimated impact.

Example promptCreate a weekly Google Ads report for April 23-30. Include: executive summary, key metrics vs targets, campaign breakdown, top 3 wins, top 3 concerns, and 5 action items with estimated impact. Format for C-suite audience.

Workflow 06

Device Performance Optimization

Device bid adjustments often remain unchanged for months despite shifting user behavior. Claude analyzes conversion rates, CPAs, and ROAS by device, compares performance across campaigns, identifies misaligned bid adjustments, and recommends specific percentage changes. Mobile conversion rates vary 40-60% by industry — automated monitoring catches these shifts quickly.

Example promptCompare device performance (mobile, desktop, tablet) across all campaigns. Calculate CPA and ROAS differences. Identify campaigns with suboptimal device bid adjustments. Recommend specific percentage changes to align with performance.

Workflow 07

Geographic Performance Analysis

Location bid adjustments based on outdated data waste 15-25% of budget in underperforming markets. Claude pulls geographic performance data, identifies locations with high spend but poor conversion rates, calculates cost-per-market efficiency, and recommends bid adjustments or exclusions. It also flags markets with strong performance that deserve increased investment.

Example promptAnalyze geographic performance by state/metro for last 60 days. Flag locations with >$500 spend and <1% conversion rate. Identify high-performing markets with low impression share. Recommend bid adjustments and budget reallocation.

Workflow 08

Keyword Expansion Research

Most Google Ads accounts capture only 30-40% of relevant search volume due to limited keyword lists. Claude analyzes top-performing search terms, extracts semantic patterns and user intent variations, researches related keywords via Google Keyword Planner integration, and suggests new keyword groups with estimated traffic and competition levels. Systematic expansion typically increases qualified traffic by 25-35%.

Example promptAnalyze top 20 converting search terms from last month. Extract intent patterns and variations. Research 50 related keywords with search volume >1000. Group by intent and recommend which ad groups to add them to.

Ryze AI — Autonomous Marketing

Skip the MCP setup — let AI optimize your Google Ads automatically

✓Automates Google, Meta + 5 more platforms
✓Handles your SEO end to end
✓Upgrades your website to convert better

2,000+

Marketers

$500M+

Ad spend

Countries

How to install Google’s official MCP server (step-by-step)

This walkthrough covers the Google official MCP server installation process, which requires Python 3.12+, Google Cloud Console access, and basic terminal familiarity. Total setup time: 15-25 minutes for first-time users. Alternative: use Ryze AI’s managed MCP connector for 2-minute setup with no technical requirements.

Step 01

Clone the official Google repository

Open terminal and run: git clone https://github.com/google-marketing-solutions/google_ads_mcp.git then cd google_ads_mcp. The repository includes the MCP server implementation, configuration templates, and documentation. Requires Git installed on your system.

git clone https://github.com/google-marketing-solutions/google_ads_mcp.git cd google_ads_mcp python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate

Step 02

Set up Google Cloud authentication

Go to Google Cloud Console > APIs & Services > Credentials. Create a new OAuth 2.0 client ID for "Desktop application." Download the JSON credentials file. Also enable the Google Ads API in the API library. Save the credentials as client_secret.json in the MCP server directory.

Step 03

Apply for Google Ads developer token

Log into your Google Ads account > Tools & Settings > API Center > Apply for developer token. Approval takes 1-2 business days for standard access. Basic tier allows 15,000 operations per day, sufficient for most MCP server usage. Test accounts can use the test developer token immediately.

Step 04

Configure environment variables

Create a .env file with your developer token, customer ID, and OAuth credentials path. The MCP server reads these variables on startup. Customer ID is your Google Ads account number without dashes (123-456-7890 becomes 1234567890).

GOOGLE_ADS_DEVELOPER_TOKEN=your_developer_token_here GOOGLE_ADS_CUSTOMER_ID=1234567890 GOOGLE_ADS_CLIENT_SECRET_PATH=./client_secret.json GOOGLE_ADS_LOGIN_CUSTOMER_ID=1234567890

Step 05

Add MCP server to Claude Desktop

Open Claude Desktop > Settings > Developer > Edit Config. Add the Google Ads MCP server configuration pointing to your local Python environment and script path. Restart Claude Desktop to activate the connection.

{ "mcpServers": { "google-ads": { "command": "python", "args": ["/path/to/google_ads_mcp/main.py"], "env": { "GOOGLE_ADS_DEVELOPER_TOKEN": "your_token", "GOOGLE_ADS_CUSTOMER_ID": "1234567890" } } } }

Step 06

Test the connection

Ask Claude: "List my accessible Google Ads accounts." If setup is correct, Claude returns your account names and customer IDs via the MCP server. Then test a basic GAQL query: "Show me campaign performance for the last 7 days." This verifies both authentication and query functionality.

How does Google’s official MCP server compare to managed solutions?

The choice between Google’s official MCP server and managed solutions like Ryze AI depends on technical comfort, customization needs, and maintenance preferences. Google’s server provides maximum flexibility through direct GAQL access but requires ongoing technical maintenance and GAQL syntax knowledge.

Feature	Google Official	Ryze AI Managed	Third-Party Bundles
Query flexibility	Full GAQL access	Pre-built functions	Structured + GAQL
Setup complexity	High (Python, OAuth, tokens)	Minimal (OAuth only)	Medium (API keys)
Maintenance	Self-managed updates	Zero maintenance	Periodic updates needed
Rate limiting	Manual handling required	Automatic management	Varies by provider
Cost	Free (API usage only)	Subscription model	Per-tool pricing

When to choose Google’s official server: You’re comfortable with Python environments, want to write custom GAQL queries, need access to specialized Google Ads resources, or require maximum data control. Best for developers building custom automation or agencies with technical teams.

When to choose managed solutions: You want immediate functionality without technical setup, prefer English-language prompts over GAQL syntax, need reliable rate limiting and error handling, or want to focus on optimization rather than infrastructure. Best for marketers, smaller teams, and rapid prototyping.

The learning curve difference is significant: Google’s server requires understanding GAQL resource relationships (campaigns.name, ad_groups.status, keywords.text) while managed solutions accept prompts like "show me underperforming campaigns." For comprehensive Claude-based Google Ads management beyond MCP, see Claude Marketing Skills Complete Guide.

What are common MCP server Google Ads setup errors?

Error 1: Invalid developer token. Using the test developer token in production or failing to get approval. Test tokens work only with test Google Ads accounts. Production accounts require an approved developer token, which takes 1-2 business days. Fix: apply for production access or use test accounts during development.

Error 2: Incorrect customer ID format. Including dashes in the customer ID (123-456-7890) instead of the numeric format (1234567890). The Google Ads API requires the 10-digit number without formatting. Fix: remove dashes and hyphens from your customer ID.

Error 3: OAuth scope mismatch. Using OAuth credentials configured for different Google APIs. Google Ads MCP requires specific scopes for the Google Ads API. Fix: create new OAuth credentials specifically for Google Ads API access in Google Cloud Console.

Error 4: Rate limit exceeded. Making too many API calls in rapid succession triggers rate limiting. The Google Ads API allows 15,000 operations per day on the standard tier. Fix: implement exponential backoff in queries or use managed solutions with built-in rate limiting.

Error 5: Python environment conflicts. Using incompatible Python versions or missing dependencies. Google’s MCP server requires Python 3.12+ and specific package versions. Fix: create a dedicated virtual environment and install exact dependency versions from requirements.txt.

Error 6: GAQL syntax mistakes. Writing invalid Google Ads Query Language queries causes API errors. GAQL requires specific resource names and field combinations. Fix: reference the Google Ads API documentation for valid resource.field combinations or use structured prompt templates.

Sarah K.

Paid Media Manager

E-commerce Agency

★★★★★

“

Setting up the MCP server took me 3 tries and 2 hours. Ryze AI’s managed connector worked immediately. Now I get Google Ads insights in seconds instead of building reports manually.”

90%

Time saved

3 min

Setup time

Real-time

Data access

Frequently asked questions

Q: What is MCP server Google Ads?

MCP server Google Ads is a bridge connecting Claude AI to Google Ads API through Model Context Protocol. It enables real-time campaign analysis, performance monitoring, and optimization recommendations without manual data exports.

Q: How long does MCP setup take?

Google’s official MCP server takes 15-25 minutes with Python/OAuth setup. Managed solutions like Ryze AI take under 3 minutes with simple OAuth authentication. Third-party bundles require 8-12 minutes.

Q: Do I need to know GAQL syntax?

Google’s official server requires GAQL (Google Ads Query Language) knowledge. Managed solutions translate English prompts to GAQL automatically, eliminating the need to learn query syntax.

Q: Can Claude make changes to my Google Ads account?

Current MCP implementations are read-only for analysis and reporting. Claude cannot execute bid changes, pause campaigns, or modify account settings. For autonomous optimization with guardrails, Ryze AI handles execution automatically.

Q: What are Google Ads API rate limits?

Standard access allows 15,000 operations per day. Basic access (default) has lower limits. Rate limiting affects MCP performance during intensive analysis. Managed solutions handle rate limiting automatically.

Q: How does this compare to manual reporting?

Manual Google Ads reporting takes 3-4 hours weekly. MCP server automation generates the same insights in 60-90 seconds with real-time data. Eliminates export/import cycles and reduces human error.