What does an attribution audit find when an app stops trusting its trackers?

What does an attribution audit find when an app stops trusting its trackers?

The uncomfortable question for mobile growth teams: how much of your "attributed" install volume is real persuasion versus trackers claiming credit for organic demand?

The case. In 2017 Uber, after a forensic review, sued one mobile ad network and cut roughly two-thirds of its mobile marketing spend — a reported $100M+ in annual budget — based on incrementality analysis. The mechanism abused was "click flooding" and "click injection": networks firing huge volumes of low-cost clicks so that last-click attribution would award them installs that would have happened organically.

What the analysis showed. When Uber paused suspect networks and watched install volume, organic installs simply rose to fill the gap. The attributed conversions were not incremental — they were demand being intercepted and relabeled. The reported lesson, repeated by their team publicly: turning off two-thirds of spend produced almost no measurable drop in genuine new users.

The nuance. This isn't only fraud. The same dynamic — a touchpoint claiming credit for an inevitable conversion — happens honestly inside last-click models every day. Fraud just industrializes the bias.

Bottom line for practitioners: the test for any channel is not "how many conversions did it touch?" but "what happens to total conversions when it goes dark?" Stagger-pause your networks and watch organic backfill. If organic rises to cover the loss, you were paying for interception. Uber's two-thirds cut held because the underlying demand never depended on those clicks.