Skip to main content
Glama
MaxGhenis

Google Ads MCP Server

by MaxGhenis
README.md
# Google Ads MCP Server

A Model Context Protocol (MCP) server for Google Ads with **read AND write capabilities**. Unlike Google's official read-only MCP server, this one lets you actually manage your campaigns.

## Features

### Read operations
- **`list_accounts`** - List all accessible Google Ads accounts
- **`execute_query`** - Run GAQL queries for campaigns, ad groups, keywords, metrics, etc.

### Write operations
- **`create_campaign`** - Create new campaigns (paused by default for safety)
- **`create_ad_group`** - Create ad groups in a campaign
- **`add_keywords`** - Add keywords to an ad group
- **`add_negative_keywords`** - Add negative keywords (campaign or ad group level)
- **`create_responsive_search_ad`** - Create RSAs with headlines/descriptions
- **`pause_campaign`** / **`enable_campaign`** - Toggle campaign status
- **`pause_ad_group`** / **`enable_ad_group`** - Toggle ad group status
- **`update_ad_group_bid`** - Update CPC bids

### Safety features
- **Dry run mode** - All write operations support `dry_run=True` to validate without executing
- **Safe defaults** - New campaigns are created PAUSED
- **Clear feedback** - All operations return success status and error messages

## Installation

### Using uv (recommended)

```bash
# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install the package
uv pip install google-ads-mcp

# Or install from source
git clone https://github.com/maxghenis/google-ads-mcp-rw
cd google-ads-mcp
uv pip install -e .
```

### Using pip

```bash
pip install google-ads-mcp
```

## Configuration

### 1. Create Google Ads API credentials

1. Go to the [Google Ads API Center](https://ads.google.com/aw/apicenter)
2. Create a developer token (or use an existing one)
3. Set up OAuth2 credentials in Google Cloud Console
4. Generate a refresh token

### 2. Create google-ads.yaml

Create a `google-ads.yaml` file with your credentials:

```yaml
developer_token: YOUR_DEVELOPER_TOKEN
client_id: YOUR_CLIENT_ID.apps.googleusercontent.com
client_secret: YOUR_CLIENT_SECRET
refresh_token: YOUR_REFRESH_TOKEN
# MCC account ID if using multiple accounts (digits only, no dashes)
login_customer_id: "1234567890"
use_proto_plus: true
```

See [google-ads.yaml.example](./google-ads.yaml.example) for a template.

### 3. Set the config path (optional)

By default, the server looks for `google-ads.yaml` in the current directory. You can specify a custom location:

```bash
export GOOGLE_ADS_CONFIG_PATH=/path/to/your/google-ads.yaml
```

## Usage with Claude Desktop

Add to your Claude Desktop config (`~/Library/Application Support/Claude/claude_desktop_config.json` on macOS):

```json
{
  "mcpServers": {
    "google-ads": {
      "command": "uv",
      "args": ["run", "--with", "google-ads-mcp", "google-ads-mcp"],
      "env": {
        "GOOGLE_ADS_CONFIG_PATH": "/path/to/your/google-ads.yaml"
      }
    }
  }
}
```

Or if installed globally:

```json
{
  "mcpServers": {
    "google-ads": {
      "command": "google-ads-mcp",
      "env": {
        "GOOGLE_ADS_CONFIG_PATH": "/path/to/your/google-ads.yaml"
      }
    }
  }
}
```

## Example workflows

### List your accounts

```
Use list_accounts to show me all my Google Ads accounts
```

### Query campaign performance

```
Run this GAQL query for customer ID 1234567890:
SELECT campaign.name, metrics.clicks, metrics.impressions, metrics.cost_micros
FROM campaign
WHERE segments.date DURING LAST_7_DAYS
ORDER BY metrics.clicks DESC
```

### Create a new campaign with ad group, keywords, and ad

```
Create a new search campaign called "Q1 Promo" with a $20/day budget,
then add an ad group "Product Keywords" with keywords like "buy widgets",
"widget sale", "best widgets", and create a responsive search ad with
these headlines: "Buy Widgets Now", "50% Off Widgets", "Free Shipping"
and descriptions: "Shop our huge selection of widgets. Free shipping on orders over $50.",
"Premium quality widgets at the best prices. Order today!"
Final URL: https://example.com/widgets
```

### Add negative keywords

```
Add these negative keywords to campaign 19638300165:
- "free"
- "cheap"
- "diy"
- "tutorial"
```

### Pause underperforming ad groups

```
First, show me all ad groups in campaign 19638300165 with their metrics.
Then pause any ad groups with CTR below 1%.
```

## API reference

### Budget and bid values

All monetary values use **micros** (1/1,000,000 of the currency unit):
- `1,000,000 micros = $1.00`
- `2,500,000 micros = $2.50`
- `10,000,000 micros = $10.00`

### Match types

Keywords support three match types:
- **BROAD** - Shows for related searches (default)
- **PHRASE** - Shows when query contains the phrase
- **EXACT** - Shows only for exact query match

### Customer IDs

Customer IDs should be digits only, no dashes:
- Correct: `1234567890`
- Incorrect: `123-456-7890`

## Development

```bash
# Clone the repo
git clone https://github.com/maxghenis/google-ads-mcp-rw
cd google-ads-mcp

# Install dev dependencies
uv pip install -e ".[dev]"

# Run tests
pytest

# Format code
black src tests
ruff check --fix src tests
```

## License

MIT License - see [LICENSE](./LICENSE) for details.

## Contributing

Contributions welcome! Please open an issue or PR.

## Related projects

- [Google's official Google Ads MCP](https://github.com/nicholasgriffintn/google-ads-mcp) - Read-only
- [MCP Protocol](https://modelcontextprotocol.io/) - The underlying protocol
- [FastMCP](https://github.com/jlowin/fastmcp) - The MCP server framework used

TDQS

A3.9/5.0

Scored across 12 tools

Disambiguation5/5

Each tool has a distinct purpose (adding keywords vs adding negatives, creating vs enabling vs pausing, etc.) with no overlapping functionality, so an agent can easily differentiate them.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern using snake_case (e.g., add_keywords, create_ad_group, pause_campaign), making the set predictable.

Tool Count5/5

12 tools is a well-scoped number for managing Google Ads campaigns, covering creation, state changes, and querying without being excessive or too sparse.

Completeness2/5

Missing crucial operations like updating campaign settings, deleting keywords or ads, and managing budgets beyond creation. The surface leaves agents with dead ends (e.g., no way to remove a mistakenly added keyword).

Maintenance

ActivityInactive
ResponsivenessNo issues