Skip to main content
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external data sources. AdAdvisor provides an MCP server so you can query your ad data directly from AI tools like Claude Code, Claude Desktop, ChatGPT, Cursor, VS Code, Gemini CLI, Codex, and Windsurf.

What does this mean in practice?

Instead of logging into AdAdvisor, navigating to the right page, and exporting data, you can just ask your AI assistant a question in natural language:
  • “What’s my ROAS for the last 7 days across all campaigns?”
  • “Which ad sets are spending the most with the lowest ROAS?”
  • “Show me the performance trend for my top 3 campaigns”
The AI assistant connects to AdAdvisor’s MCP server, runs the query, and gives you the answer right in your editor or terminal.

How authentication works

There are two ways to authenticate with the MCP server:

OAuth (browser login)

Most MCP clients support OAuth. When you connect for the first time, a browser window opens where you log in to AdAdvisor. The client receives a token automatically and handles refreshing it for you. No API key needed. This works in both desktop apps (like Claude Desktop and ChatGPT) and terminal tools (like Claude Code and Gemini CLI).

API Key

If you prefer not to use OAuth, or if your setup doesn’t support it, you can create an API key in AdAdvisor and pass it as a header in your client config. This is also useful for shared or automated setups where a browser login isn’t practical.
API keys are scoped to your organization. Anyone on your team with an API key can query the organization’s ad data.

What data is available through MCP?

The MCP server exposes the same data you see in AdAdvisor:
  • Campaign, ad set, and ad performance metrics
  • Recommendations and their statuses
  • Business metrics and configuration
  • Account-level insights

Getting started

Last modified on February 27, 2026