Brian Morse – Head of Collective, Locality
Prasad Joglegkar – Founder and CEO, Deben Media Corp.
Brian Morse (Locality) and Prasad Joglegkar (Deben Media) describe their data-driven approach to improving cross-platform audience reach measurement by correcting the widely used—but flawed—assumption of random duplication. They analyze commercially available datasets across multiple markets and platforms, including broadcast TV, cable, streaming/OTT and online video, to demonstrate that audience overlap between publishers is systematically undercounted when using random duplication formulas.
Download Presentation
Member Only Access
Brian Morse – Head of Collective, Locality
Prasad Joglegkar – Founder and CEO, Deben Media Corp.
Brian Morse (Locality) and Prasad Joglegkar (Deben Media) describe their data-driven approach to improving cross-platform audience reach measurement by correcting the widely used—but flawed—assumption of random duplication. They analyze commercially available datasets across multiple markets and platforms, including broadcast TV, cable, streaming/OTT and online video, to demonstrate that audience overlap between publishers is systematically undercounted when using random duplication formulas.
Their empirical research examined thousands of network pairs (e.g., 6,000+ combinations), compared estimated vs. actual overlap using multiple currency datasets (panel-based, ACR, etc.) and normalized results to a common population baseline in a DMA-level universe. The findings consistently show that duplication follows predictable behavioral patterns that are aligned with established statistical models like the duplication of viewing law and negative binomial distribution, rather than randomness.
Building on these observations, the researchers proposed practical modeling recommendations that materially improve media allocation decisions, as illustrated by a pharmaceutical advertiser case where correcting duplication assumptions increased effective reach from 48% to 57%. The implications span multiple industry verticals where inaccurate overlap estimates can lead to over-frequency, inefficient spend and misleading campaign reporting. The research showed that accurate reach modeling must be grounded in observed human viewing behavior across linear and digital ecosystems, rather than simplistic assumptions.
Key takeaways:
- Empirical analysis shows duplication follows established statistical patterns, not independence, showing that duplication behavior is not random. Real-world audience overlap is systematic and consistently undercounted across datasets and platforms.
- A simple correction factor (D ≈ 1.3) can reduce error by 40–47%, while more advanced models further improve precision in cross-platform planning.
- Accurate de-duping materially changes media decisions. Inaccurate duplication estimates have major business consequences—driving over-frequency, inefficient media spend and misaligned campaign outcomes across industries like advertising and pharma.
Watch the Presentation
Download Presentation
Member Only Access
