Free guide for performance marketers

How to Connect Claude to Your Google Ads + Meta Ads Data

Stop exporting CSVs. Ask your ad data questions directly and get answers in plain English — a step-by-step setup guide with 5 copy-paste cross-channel prompts.

Read the guide below

Built for agencies and in-house marketing teams.

Backed by

StanfordStartup Incubated

GoogleStartup Program

What this setup unlocks

Once connected via MCP, Claude becomes your cross-channel analyst.

Live Campaign Data

Pull live campaign metrics directly from Google Ads and Meta without ever exporting another CSV.

Spot Budget Leaks

Compare ROAS and CPAs across both channels in a single conversation to find exactly where budget is bleeding.

5 Cross-Channel Prompts

Get our exact prompts to run creative fatigue checks, ROAS discrepancy analyses, and weekly summaries.

Inside the guide

Google officially open-sourced their MCP server and we can access it directly from GitHub. It's read-only and built for diagnostics and analytics — exactly what we need. For Meta, we'll use the widely-adopted open-source server by Pipeboard.

Requirements: Claude Desktop + Claude Pro plan ($20/month) + ~30 minutes of setup.

Part 1: Connect Google Ads to Claude

GitHub: googleads/google-ads-mcp

Step 1: Get your Google Ads API credentials

Go to Google Cloud Console and create a new project
Enable the Google Ads API
Create OAuth 2.0 credentials (Desktop App type)
Apply for a Developer Token in your Google Ads account under Tools → API Center

A basic access token is sufficient for read-only use.

Step 2: Install the MCP server

bash

git clone https://github.com/googleads/google-ads-mcp
cd google-ads-mcp
pip install -r requirements.txt

Step 3: Configure Claude Desktop

Open your Claude Desktop config file: ~/Library/Application Support/Claude/claude_desktop_config.json and add the following:

claude_desktop_config.json

{
  "mcpServers": {
    "google-ads": {
      "command": "python",
      "args": ["/path/to/google-ads-mcp/server.py"],
      "env": {
        "GOOGLE_ADS_DEVELOPER_TOKEN": "your_developer_token",
        "GOOGLE_ADS_CLIENT_ID": "your_client_id",
        "GOOGLE_ADS_CLIENT_SECRET": "your_client_secret",
        "GOOGLE_ADS_REFRESH_TOKEN": "your_refresh_token",
        "GOOGLE_ADS_LOGIN_CUSTOMER_ID": "your_customer_id"
      }
    }
  }
}

Step 4: Restart Claude Desktop and test

Restart the app and try:

“Show me my top 5 Google Ads campaigns by spend in the last 14 days”

If you see campaign data returned, you're connected.

Part 2: Connect Meta Ads to Claude

GitHub: pipeboard-co/meta-ads-mcp

Option A — Remote MCP (Recommended, no local setup)

The fastest path:

Go to claude.ai/settings/integrations
Click Add Integration
Add URL: https://mcp.pipeboard.co/meta-ads-mcp
Log in with your Meta account
Done — no config files, no local installation

Option B — Local install

bash

pip install meta-ads-mcp

Then add to your Claude Desktop config:

claude_desktop_config.json

{
  "mcpServers": {
    "meta-ads": {
      "command": "uvx",
      "args": ["meta-ads-mcp"],
      "env": {
        "META_ACCESS_TOKEN": "your_token_here"
      }
    }
  }
}

To get your Meta access token:

Go to Meta Business Suite
Business Settings → Users → System Users
Create a system user → Generate token
Required permissions: ads_read, ads_management

Step 3 — Test the connection

“Show me my active Meta ad sets and their current CPMs”

If you get campaign data back, both platforms are live.

Skip the manual setup

Want all this without the config files?

Connect your Google, Meta, TikTok, LinkedIn, and Shopify accounts once. third i surfaces cross-channel insights every morning — automatically — and makes the fix once you approve it. No prompts, no APIs, no CSVs.

Start your free trial

Part 3: 5 cross-channel prompts to run today

This is where it gets useful. Each prompt works because Claude now has access to both your Google and Meta data in the same conversation. Copy, paste, and adjust the bracketed values.

Prompt 1: Budget Leak Detector

Google Ads + Meta

Look at my Google Ads and Meta Ads spend for the last 14 days.

For each channel:
- What is the total spend?
- What is the CPL or CPA?
- Is the trend improving or worsening week-over-week?

Flag any channel where CPL has increased more than 20% in the last 7 days.
Tell me where budget is leaking and which channel deserves more or less investment right now.

Prompt 2: ROAS Discrepancy Check

Google Ads + Meta

My Meta dashboard is showing a strong ROAS but overall revenue hasn't grown proportionally.
Pull my Meta ROAS and Google ROAS for the last 30 days.

Cross-reference the spend and conversion volumes on both platforms.

Give me 3 possible explanations for the discrepancy -
could this be audience overlap, attribution window mismatch,
or assisted conversions being counted twice?

Prompt 3: Creative Fatigue Cross-Check

Google Ads + Meta

On Meta: identify any ad creatives with frequency above 2.5 in the last 7 days.
On Google: are any of the same audience segments being reached via Display or Demand Gen?

Flag where the same user is likely seeing our ads on both platforms at high frequency.
Recommend which creatives to refresh first based on spend and fatigue signals.

Prompt 4: Budget Reallocation Signal

Google Ads + Meta

My total daily budget across Google and Meta is $[YOUR_BUDGET].
Based on the last 14 days of performance:

- Which campaigns on each platform are hitting below-target CPA?
- Which are performing above target and could absorb more budget?

Recommend a reallocation plan.
Apply the rule: don't move more than 20% of any campaign's budget in one day to avoid resetting learning phases.

Prompt 5: Weekly Cross-Channel Narrative

Google Ads + Meta

Summarize last week's performance across Google Ads and Meta Ads in plain language.
Structure it as:

1. What worked (top performing campaigns, creatives, audiences)
2. What didn't (underperformers, wasted spend, anomalies)
3. What to do differently this week (3 specific actions)

Write it so I can paste this directly into a client report or a Monday team standup.

Done-for-you

See it without the setup

Don't want to wrangle config files and API tokens? Tell us about your stack and we'll show you the cross-channel insights third i surfaces — connected once, no CSVs, no prompts. First insights within 24 hours of connecting your accounts.

Connect Google, Meta, TikTok, LinkedIn & Shopify once
Daily cross-channel insights, automatically
Paid, organic, content and revenue in one workspace