Substitution among Social Media Platforms: Evidence from App Tracking Panel Data

Abstract

This thesis explores a novel approach to competitive intelligence in the social media ecosystem by leveraging external mobile panel data to study substitution dynamics. It focuses on contextspecific behavioral patterns to identify which platforms compete for user attention in given situations. Using mobile app session data from April 2023 for approximately 5,000 users, the analysis segments usage into three behavioral contexts – morning, evening, and at-home sessions – and characterizes user-app interactions through descriptive statistics. K-means clustering is applied to identify archetypes of usage behavior across these contexts, revealing distinct patterns such as quick-check habits, deep content consumption, and intensive texting. By comparing app usage profiles across contexts, the study uncovers shifts in how and when platforms are used, highlighting subtle substitution dynamics. To validate the findings, the study analyzes app usage during service outages, testing if potential substitutes see increased engagement when a competing platform is unavailable. These insights offer a richer, contextaware framework for product managers to uncover indirect competition and tailor platform strategies to specific user behaviors. Limitations include reliance on behavioral data without content-level detail, mobile-only focus, and demographic skew in the panel.