MMM Pilot™

Getting Started

This guide takes you from zero to your first Marketing Mix Model results — step by step, no data science background required. MMM Pilot runs your models on Robyn, Meta’s open-source MMM framework, so every analysis is backed by a methodology trusted by the wider MMM community.

Step 1: Create an Account

  1. Navigate to the MMM Pilot login page.
  2. Click Sign Up.
  3. Enter your email address and create a password.
  4. Select your experience level — this helps the platform tailor suggestions to your background.
  5. Confirm your email if prompted.
  6. You are redirected to the main dashboard.

Your personal workspace is created automatically. You are ready to build your first project.

Step 2: Create a Project

  1. From the dashboard, click the New Project button.
  2. The Create New Project dialog opens. Fill in:
    • Project Name (required) — A descriptive name, e.g., “Q1 2025 Marketing Analysis”.
    • Country — Your primary market. This affects external data sources such as FRED economic indicators.
    • Currency — The currency for spend data (defaults to your profile preference).
    • Industry — Helps the AI generate better data suggestions.
    • Product/Service Categories — Comma-separated product types.
    • Main Competitors — Competitor URLs or names, one per line.
    • Additional Context — Any other relevant business information.
  3. Click Create Project.
  4. You are redirected to the project’s Data tab.

Step 3: Import Your Data

You need at least one data source containing a date column and one business outcome variable to run a model.

Option A: Upload a CSV

  1. On the Data tab, click Add Data → CSV Upload.
  2. Select or drag a CSV file into the upload area.
  3. The platform detects the date column and data granularity automatically.
  4. Preview the data and confirm.
  5. The data source appears in the project’s data list.

Option B: Connect a Platform

  1. Click Add Data and choose Google Ads, Meta Ads, or Google Analytics 4.
  2. Follow the OAuth authentication flow.
  3. Select the account or property and import data for your desired date range.

See the individual integration guides for detailed steps on each platform.

Option C: Discover External Data

  1. Click Add Data → Discover FRED or Discover Google Trends.
  2. Review the AI-suggested indicators or search terms.
  3. Select the ones most relevant to your business and import them.

Step 4: Configure Field Types

Before running an analysis, each data field needs a role assignment so the model knows how to treat it.

  1. Open the data source preview cards on the Data tab.
  2. For each field, set:
    • Field Alias — A human-readable display name.
    • Field Type — Dependent variable, paid media, organic, or context.
  3. Resolve any field name conflicts if multiple data sources share column names.

Step 5: Run Your First Analysis

  1. Click the New Run button on the project page.
  2. The Run Configuration Dialog opens, showing:
    • All available fields grouped by type.
    • AI Suggest buttons for automatic field role assignment and hyperparameter recommendations.
    • Time window selection.
    • Model parameters (iterations and trials).
  3. Give the run a descriptive name.
  4. Click Start Run.
  5. The estimated credit cost is displayed in the dialog. Credits are deducted upon successful run completion. The run appears in the sidebar with a “Processing” status — you can close the browser and come back when it’s done.

Step 6: Review Your Results

When the run completes:

  1. Click on the run in the sidebar to open the Run Results page.
  2. Review the AI Evaluation Report at the top for an overall quality assessment of your model.
  3. Browse the artifact categories:
    • Model Evaluation — Pareto front, one-pager summary.
    • Plots — Spend Efficiency curves, Contribution Breakdown, waterfall chart.
    • Solution Files — Downloadable CSVs with detailed Contribution Breakdown data.
  4. If the AI suggests refined hyperparameters, use the one-click button to start an improved iteration.

What to Do Next

  • Iterate — Use AI-suggested hyperparameters to refine your model and improve fit.
  • Add more data — Bring in FRED indicators or Google Trends to improve explanatory power.
  • Share results — Invite team members or stakeholders to view your project.
  • Explore guides — Dive into the individual guides for each feature and integration.