📊 simba-mcp - Connect AI tools to marketing models

Simba-mcp enables your AI assistants like Claude or Cursor to interact with marketing mix models. You use natural language to upload your data, create performance models, optimize your marketing budgets, and test future scenarios. The tool bridges the gap between complex statistical analysis and daily marketing operations.

🚀 What this tool does

Marketing analytics often require specialized software and complex manual processes. This project changes that dynamic by allowing AI models to handle the heavy lifting. Once you connect your data, the system builds a Bayesian hierarchy to calculate your media performance. You ask questions about ROI or budget allocation, and the system answers based on your unique data.

📋 System requirements

• Windows 10 or Windows 11
• 8 GB RAM minimum (16 GB recommended)
• Stable internet connection
• Access to an AI assistant that supports the Model Context Protocol (MCP)

📥 Downloading the software

Visit this page to download the package. Choose the latest release version that matches your Windows architecture. We recommend the standard installer for most users.

⚙️ Installation steps

1. Locate the downloaded file in your computer folders.
2. Double-click the file to start the setup process.
3. Follow the on-screen instructions to extract the files.
4. Keep note of the folder location where you save the files.
5. Grant the system permission if Windows asks for verification.

🖱️ Connecting to your AI assistant

The connection process allows your AI tool to send instructions to the software.

1. Open your chosen AI assistant settings or configuration menu.
2. Look for the "MCP Servers" or "Tools" section.
3. Select the "Add New Server" option.
4. Copy the file path of your simba-mcp installation.
5. Save the configuration.
6. The AI assistant should now recognize the tools provided by the software.

📈 Analyzing your data

Preparation of your data helps the AI generate accurate models. Organize your marketing spend and outcome data in a CSV file. Ensure the following columns exist:

• Date (YYYY-MM-DD)
• Media Spend per Channel
• Total Revenue or Conversions

Upload this file when prompted by your AI assistant. The software processes these inputs to create a baseline model. You can then ask the AI to perform specific tasks, such as finding the optimal budget split for the next quarter.

🧩 Optimizing budgets

Budget optimization works by running multiple scenarios. The system simulates outcomes based on the trends found in your data. You can ask for a recommendation based on a specific budget cap or a target revenue goal. The AI communicates these results in plain text, explaining which channels show the most growth potential.

🛠 Troubleshooting common issues

If the AI fails to connect, close both your AI assistant and the background server process. Restart the assistant first to refresh the connection. Ensure your data file uses the correct format and has no empty or missing rows. If a model fails to build, check that your data covers a sufficient date range, typically at least 12 months for reliable results.

🏗 Understanding the technology

This tool uses the Model Context Protocol to manage interactions. It runs local statistical routines to ensure your data stays on your machine during the calculation phase. The Bayesian methods focus on identifying patterns in media attribution that standard averages often miss. This approach minimizes error and identifies the contribution of offline and online media reach.

📘 Helpful resources

You can find further details about Bayesian modeling in public marketing analytics forums. If you encounter bugs or want to request features, check the main repository page for instructions on how to contribute or report information. Use the issue tracker to log any technical errors. The community maintains documentation for advanced users who wish to customize the underlying statistical parameters. Always check for updates to ensure compatibility with your current AI assistant version.