Course contentsShow
← Data Science
Lesson 1638 of 2,14539. Business Metrics and KPI DesignFree lesson

Direct vs Indirect Attribution

Distinguishing between metrics directly caused by an action versus those influenced indirectly through intermediary factors.

Direct vs Indirect Attribution

You'll learn: How to distinguish between metrics directly caused by an action and those influenced indirectly through one or more intermediary steps.

The Core Distinction

When attributing changes in business metrics to specific actions, you need to separate direct attribution—where an action immediately affects the metric—from indirect attribution, where an action influences the metric through an intermediary factor.

Think of it like watering a plant. Direct attribution would be measuring soil moisture after watering—the water directly changes moisture levels. Indirect attribution would be measuring plant growth: watering affects soil moisture, which affects nutrient uptake, which eventually influences growth. The chain is longer and involves multiple causal steps.

Why This Matters for Metric Design

In your metric tree from the previous lessons, direct attribution applies to metrics immediately downstream of an action. For example:

  • Launching a discount → directly increases order volume
  • Adding a feature → directly increases feature usage

Indirect attribution involves secondary or tertiary effects:

  • Discount → order volume → repeat purchase rate → customer lifetime value
  • Feature launch → feature usage → session length → retention → revenue

The further you move from the action, the more confounding factors enter the picture. Other teams' initiatives, seasonality, external market forces—all can muddy the attribution.

Practical Implications

Direct attribution is cleaner and easier to measure in A/B tests. Indirect attribution requires more sophisticated causal inference techniques and longer observation windows. When setting OKRs, prefer directly attributable metrics for near-term key results, while using indirectly influenced metrics (like your North Star) for longer-term objectives.

Key Takeaway: Direct attribution links actions to immediate metric changes; indirect attribution traces influence through intermediary factors—understanding this distinction helps you set realistic expectations for measurement timelines and causal confidence.