Advertising fingerprinting and bad incentives: “The system is broken by design”

In this new episode of the Mobile Dev Memo podcast, I speak with Maor Sadra, the CEO of INCRMNTAL, about probabilistic attribution for mobile advertising and why bad incentives prevent advertisers from rejecting it. We also talk about Media Mix Models and Facebook’s efforts to drive adoption for Robyn, its open-sourced measurement product, and advertising budget diversification post-ATT.

As always, the Mobile Dev Memo podcast is available on:

Fingerprinting on mobile is a fairly controversial topic at the moment. I’ve written about the practice extensively, but it’s worth noting that my general thesis around Apple’s handing of fingerprinting has thus far proven to be incorrect. I believed that Apple would police the use of device parameters for the purposes of install attribution — since Apple has very plainly stated that using such parameters contravenes ATT policy — but thus far, it hasn’t. And given the amount of time that has lapsed since the introduction of ATT, it seems unlikely that Apple will police the practice of probabilistic install attribution using device parameters (read: fingerprinting), at least through the mechanisms available to it, as I outline here.

This most recent earnings season presented stock analysts with the opportunity to ask ad tech CEOs about the long-term viability of probabilistic attribution using device parameters. But, as far as I can tell, none did. Does Apple actually care about fingerprinting? Perhaps it doesn’t — fingerprinting doesn’t allow for the “events stream,” as I’ve called the process of observing conversions to build user profiles, and so it’s possible that Apple continues to look the other way as ad tech vendors perpetrate it.