Project Overview

  • The goal of this work was to reduce CAQ and even if it was expensive, to make sure we get the best quality of users for the buck. This was done alongside marketing team and business analytics team
  • The idea behind pivoting for quality of users is based on the fact that major profit in the long term would be from a small portion of these customers rather than from everyone
  • This involved working with Google Ad Analytics data, FB ad data, Affiliates data through Appsflyer attribution and other user identifying information from priors
  • In addition to the increase in quality and reduction in CAQ, there was also a risk related cost to be mitigated by preventing onboarding of bad actors
  • This bad actor identification involved multiple signals from varied data sources such as EPFO, PAN, location etc.

Skills

Knowing what and who we want

This project revealed the need to optimize who we need to target through our Acquisition channels and have a clear definition and need of why. This required bridging user characteristic pre and post acquisition

Working around limited control knobs

While we may have several markers and control over them post onboarding, we may not necessarily have controls over the markers through the acquisition platforms. It was required to work around this challenge and create alternate control methods

Multi-level controls

Often there is no single layer of control across the different markers. Hence this efficient targeting requires several separate analysis and modeling, that all come back together to tie neatly to the bigger picture