Key Metric Drivers Report

Overview

The Key Metric Drivers report identifies which metrics have the strongest influence on your primary business metric. By analyzing statistical relationships between metrics, this report helps you understand which factors contribute most significantly to your key performance indicators (KPIs) and guides data-driven decision-making.

How to Create This Report

1. Add metrics to your board

   • Add the main metric you want to analyze
   • Add the potential driver metrics you want to test

2. Connect your metrics

   • Draw connections from each potential driver metric to your main metric

3. Generate the report

   • Select your main metric on the board
   • Click on the “Report” button in the popup menu
   • Choose “Key Metric Drivers” from the report options

What It Measures

1. Overall Model Effectiveness

   • The R-Squared value shows what percentage of changes in your main metric can be explained by the dependent metrics you’ve selected.
   • A higher R-Squared value (closer to 100%) indicates that the model effectively captures the key relationships between your metrics.

2. Metric Contributions & Impact

   • The report shows the percentage contribution of each dependent metric, clearly indicating their relative impact on your main metric.
   • Both raw and standardized coefficients are presented, allowing you to compare the influence of different metrics regardless of their scale.

3. Correlations & Significance

   • Correlation coefficients measure the strength and direction of relationships between your metrics.
   • Statistical significance indicators help you determine which relationships are meaningful for decision-making versus those that might be due to chance.

4. Time Series Contributions

   • The report analyzes how each metric’s impact changes over time, helping you identify trends, seasonal patterns, or shifts in metric relationships.
   • This temporal view helps you understand if certain metrics are becoming more or less influential.

Statistical Considerations

Multicollinearity

• The report uses Variance Inflation Factors (VIF) to detect when metrics are too closely related to each other.
• High VIF values indicate that certain metrics may be redundant, potentially distorting your understanding of what’s truly driving your main metric.

Non-Significant Variables

• The report identifies metrics with high correlation but non-significant statistical impact, which may indicate confounding effects.
• Metrics with high p-values (typically above 0.05) might not be reliable drivers and should be evaluated further.