1. Accessing the MMM Results Page 0:09
Navigate to the MMM Results page in two ways:
Go under 'More' and click on 'Marketing Mixed Modeling'.
Customize the navigation bar for easy access.
Alternatively, access it from the Model Settings by clicking 'View Dashboard'.
2. Overview of Model Settings 0:31
The model focuses on optimizing for New Customer Revenue.
Key metrics:
Optimized vs. Expected New Customer Revenue:
Expected: Simulation of revenue based on current spend.
Optimized: Maximum revenue potential based on simulations.
3. Understanding Revenue Predictions 1:36
Example prediction:
Spending $622,000 this week.
Expected revenue: $583,000 with a 0.94 new customer ROAS.
Optimized potential: Increase ROAS by 12% to 1.05, leading to additional revenue.
4. Custom Spend Integration 2:39
Ability to add custom spend (e.g., mailers, TV) to the MMM model.
Documentation available for integrating custom spend into TripleWhale.
5. Daily Spend Recommendations 3:37
Expected daily spend on Google Ads vs. optimized recommendations:
Current expected: $13,576.
Optimized recommendation: $11,428 (16% reduction).
6. Campaign Type Analysis 4:49
Model operates on segmented campaign types (e.g., Performance Max, Brand).
Recommendations are based on historical performance of each segment.
7. Marginal New Customer ROAS 6:36
Definition: Expected return on the next dollar spent.
Example:
Performance Max: Next dollar yields higher ROAS.
Retargeting: Lower expected return, indicating saturation.
8. Pacing Towards Revenue Goals 8:10
Comparison of predicted vs. actual revenue:
Blue line: Optimized revenue goal.
Green line: Actual revenue.
9. Model Fit and Historical Performance 9:20
Evaluation of model accuracy:
Metrics include MAPE, CRPS, Pearson, and R-squared.
10. Training vs. Testing of the Model 13:14
Training Phase: Model learns from both spend and revenue data.
Testing Phase: Model predicts revenue based solely on spend data.
11. Summary of Recommendations and Insights 15:38
Top-line recommendations:
Confidence score, budget recommendations, pacing towards predictions.
Breakdown of performance by channel and campaign type.