[alexeichemenda]

What's important to highlight here is that marketing teams at companies such as Uber are incentivized to pus vendors to drive fraud. Specific example: Uber works with vendors A, B, C. Marketing team is incentivized to reach a cost per sign-up of $X (Let's say $50 for the sake of this example). Vendor A, running fraud, delivers sign-ups @ 45$ each. Vendor B, clean, delivers them at $55 each. Vendor C @ $65 each. The new baseline from an exec standpoint is $45 each, and every vendor that doesn't deliver at that level is cut. Repeat with multiple vendors.

The solution to this problem is incrementality measurement at the channel level. Every time you scale with a recently onboarded vendor, measure baseline of ALL conversions happening on your app. If this baseline doesn't move, cut the vendor. I say scale and not launch because upon launh, there won't be a visible impact on the global conversions. To be able to spot this spike from baseline, pick a small market than "worldwide". For ex, pick "California", let the new vendor scale in California, and measure spike in California.

[alexeichemenda]

I'll also mention: this is a problem that goes up every ladder. Marketing individual contributor wants to show good performance to their manager, so delivering rides for cheap is good. Marketing manager is in the same boat with the CMO. CMO -> Board -> VC (VC will be happy to see great efficiency on the customer acquisition side). VC -> LPs (LPs will be excited about customer acquisition efficiency). A limited number of people are actually deeply concerned with this, and that's why it's taking so long for the top KPIs to change from "cost per action" to "incremental impact and incremental ROI".