How to Run Meridian MMM on Windsor.ai Data (A Step-by-Step Guide for Non-Techies)

This guide walks you through the steps required to run the Meridian MMM (Marketing Mix Modeling) tool using Windsor.ai data via a pre-built Docker container image.

Prerequisites:

• Windsor.ai API key (get yours from https://onboard.windsor.ai/).
• Docker Desktop installed and running on your machine.
• To get the most accurate results, various ad data sources (Google Ads, Facebook Ads, Bing Ads, etc.) and Google Analytics 4 should be connected in your Windsor.ai account.

Step 1: Install Docker Desktop

If you don’t have Docker Desktop installed yet:

1. Visit https://www.docker.com/products/docker-desktop/.
2. Download and install the app for your operating system (macOS, Windows, or Linux).
3. Launch Docker Desktop and ensure it’s running properly.

Step 2: Pull the Windsor-Meridian Docker Image

Open your system’s terminal and run:

docker pull windsorai/windsor-meridian

This command automatically pulls the Docker image from Docker Hub.

Step 3: Verify the image in Docker Desktop

To verify that the required image is pulled:

1. Open Docker Desktop.
2. Navigate to the Images tab.
3. You should see windsorai/windsor-meridian listed there.

Step 4: Navigate to your working directory

Before running the Docker container, navigate to the folder where you want the model’s output files to be saved.

You can create a new folder on your Desktop and move into it using the cd (change directory) command. For example:

cd {your_path}/{your_folder_name}

Note: This step is essential because we’ll mount this folder into the container, ensuring that any output files generated by the model are saved directly to your local machine in the specified folder.

Step 5: Run the Docker Container

Now run the following command in your terminal:

docker run --rm -it -v "$(pwd):/app/output" windsorai/windsor-meridian

Explanation of flags:

• –rm: Automatically removes the container after it exits.
• -it: Runs in interactive mode (so you can provide inputs).
• -v “$(pwd):/app/output”: Mounts your current directory as the output folder inside the container.

Step 6: Provide inputs to the model

Once the container starts running, it will ask you to provide the following values in your terminal:

1. API Key: Enter your Windsor.ai API key when prompted.
2. Select KPI: Choose between Revenue or Conversions as your model’s Key Performance Indicator.
3. If Conversions is selected: You will see a list of available events assigned to your account. Enter the event(s) you want to use for the conversion output.

Step 7: View the output files

Once the model runs successfully, you will see the “Process Finished! Summary output file is created.” message in your terminal, and it will generate the following files in your chosen directory:

Open the summary_output.html file to explore your MMM results visually.

Summary output example

Cheers!

You’ve successfully run Meridian MMM on Windsor.ai data using Docker. You can now review the results, fine-tune your inputs, and re-run the model as needed to gain deeper insights into the performance of your paid vs. organic marketing channels.