Incrementality Testing

Why Incrementality Matters

Attribution tells you who gets credit, but not whether your ad actually caused the conversion. Many attributed conversions would have happened anyway through organic channels. Incrementality testing reveals your true marketing impact.

Testing Methodologies

1. Geo Holdout Tests

Split similar geographic regions into test and control groups. Run ads in test regions, hold back in control regions, and measure the difference in conversions.

2. Audience Holdout Tests

Within a platform, exclude a random subset of your target audience from seeing ads. Compare conversion rates between exposed and unexposed groups.

3. Ghost Ads

Track when your ad would have been shown in an auction but wasn't. Compare users who saw your ad vs. those who saw a competitor ad in the same placement.

What to Test

• Channel incrementality: Is this network truly driving lift?
• Budget levels: At what point do diminishing returns kick in?
• Creative impact: Which creatives drive true incremental value?
• Audience segments: Which users are most incremental?
• Retargeting: How much lift does remarketing actually provide?

Implementing Incrementality

• Start with your largest channels first
• Run tests for 2-4 weeks minimum for significance
• Account for seasonality in test design
• Use multiple test methods to validate results
• Reallocate budget based on iROAS, not attributed ROAS