The Mobile User Acquisition Show
šAre You Overpaying Your Ad Network by 30-40%? Decoding SKAN vs. probabilistic billingš§
Podcast summary created with Snipd AI
Quick takeaways
- Working with ad networks that use probabilistic modeling for attribution and reporting while relying on SKAN data for billing can result in a 30-40% increase in actual spend compared to reported spend.
- To address the discrepancy in ad network metrics, advertisers can either take an aggressive approach by considering billable spend as the source of truth and adding a multiplier for missed conversions, or a conservative approach by relying on scan conversions as the source of truth and disregarding probabilistic conversions, ensuring easier comparison across channels and more credible metrics.
Deep dives
The Pitfall of Using Probabilistic Modeling for Ad Network Attribution
One main idea discussed in the podcast is the potential pitfall of working with ad networks that use probabilistic modeling for attribution and reporting. The speaker highlights that while these networks may use probabilistic models for optimization and reporting, they rely on scan data for billing. This discrepancy can lead to a significant difference between the spend accounted for in daily and weekly reports and the actual bill from ad networks, with a potential increase of 30 to 40%. This gap can result in significant financial loss if not carefully monitored and accounted for, especially if there is a disconnect between the finance team and the media buying team. The example of the difference in reported spend between probabilistic and billable according to scan data is provided to illustrate the magnitude of the issue.
Remember Everything You Learn from Podcasts
