Why Platform-Reported ROAS Overstates True Incrementality

Independent geo-holdout studies consistently find that platform-reported ROAS overstates true incremental ROAS by 20-50%, largely because platform attribution credits conversions that would have happened anyway (branded search, retargeting of already-converting users).

• Retargeting campaigns show the largest gap between platform-attributed ROAS and true incremental lift
• Geo-holdout tests are the most reliable (if operationally heavier) way to measure true incrementality at the channel level
• Brands should maintain both a platform-reported ROAS (for optimization) and a periodic incrementality-adjusted ROAS (for budget decisions)
• Incrementality testing requires meaningful budget and sample size — small brands may need quarterly rather than always-on testing

• • Justifying a shift in budget away from over-credited retargeting spend
• • Building an executive-level 'true ROAS' metric distinct from platform dashboards