Getting started with Cassandra
      Back to home
      1. Knowledge base
      2. Getting started with Cassandra

      Getting started with Cassandra

      Welcome to Cassandra! This guide will help you set up your account, integrate data, build your first marketing mix model (MMM), and optimize it over time.

      Step 1: Account Setup

      1. Sign Up & Log In
        • If you received an invitation, click the link in the email to create your password.
        • If signing up manually, visit Cassandra.app and register.
      2. User Roles & Permissions
        • Premium User: Full access, can build models and integrate data.
        • Analyst: Can view models and reports but cannot modify them.
        • View-Only: Read access only.
      3. Invite Team Members
        • Navigate to Settings > Team Management to invite colleagues.
        • Assign appropriate roles based on their responsibilities.

      Step 2: Data Integration

      1. Required Data for MMM:
        • Marketing Spend Data: Google Ads, Meta, TV, CRM, etc.
        • Sales/Conversion Data: New customers, transactions, revenue.
        • External Factors: Seasonality, promotions, economic indicators.
      2. Upload Your Data:
        • Go to Data Integration > Upload CSV (for manual uploads).
        • Connect to platforms like Google Ads, Facebook, or CRM using the API integrations.
        • Ensure data is formatted correctly:
          • Date format: YYYY-MM-DD
          • Spend values should be numeric (no currency symbols).
          • Granularity: Weekly data is recommended.
          • At least 4-6 weeks of historical data is required for new channels.
      3. Automate Data Refreshing:
        • Cassandra supports automated data pipelines via APIs to keep your model updated without manual uploads.

      Step 3: Building Your First Model

      1. Create a New Model:
        • Navigate to Modeling Dashboard > Create New Model.
        • Select your dataset and define modeling parameters.
      2. Map Variables:
        • KPI (Key Performance Indicator): Revenue, new customers, or conversions.
        • Marketing Channels: Paid media sources (Google Ads, Meta, TV, etc.).
        • Contextual Variables: External factors like promotions and seasonality.
      3. Define Your Modeling Window:
        • Recommended: Last 2 years of data.
        • Avoid data shorter than 10 weeks per channel.
      4. Run & Evaluate the Model:
        • Click Run Model to generate results.
        • Review Accuracy Score, Confidence Intervals, and Error Rates.
        • Cassandra highlights uncertainty in model results (e.g., low-spend channels may have wider confidence intervals).

      Step 4: Interpreting Results & Optimization

      1. Understanding Model Insights:
        • Channel Contributions: How much each marketing channel drives conversions.
        • ROI Analysis: Which campaigns are most effective?
        • Budget Recommendations: Suggested allocation for optimal performance.
      2. Improving Model Accuracy:
        • Remove variables with low spend (<3% of budget).
        • Add external factors like seasonality, promotions, and website changes.
        • Test different modeling windows for better fit.
        • Use multi-touch attribution to compare Cassandra’s incremental results with platform-reported conversions.
      3. Incrementality Testing & Validation:
        • Run GeoLift or Conversion Lift experiments to validate model assumptions.
        • Use findings to refine and calibrate future models.
        • Cassandra provides calibration options to adjust model weights based on experimental results.

      Step 5: Updating & Refreshing Models

      1. Refreshing Data:
        • Upload updated spend & conversion data monthly.
        • Navigate to Models > Refresh Model to apply new data.
        • Cassandra will automatically adjust forecasts and detect seasonality trends.
      2. Comparing Model Iterations:
        • Track improvements over time.
        • Adjust marketing investments based on new findings.
        • Evaluate confidence intervals to determine uncertainty levels.
      3. Budget Optimization & Experimentation:
        • Use the Budget Allocator to adjust spend based on model recommendations.
        • Test alternative spend levels and see forecasted revenue impacts.
        • Run incrementality experiments to validate high-uncertainty channels.
      4. Exporting & Sharing Results:
        • Download reports in CSV, PDF, or share via Cassandra’s dashboard.

      Need Help?

      • Live Chat: Available within Cassandra’s platform
      • Slack Integration: Available for enterprise clients