Stages of User Engagement

Analysis of Social Commerce Platforms with Navigational Clickstream Data

Kumar, Salo & Li — International Journal of Electronic Commerce (2019)

Introduction & Research Gap

The Evolution of Platforms

  • Brand communities [1, 51] \(\to\) Social commerce platforms [6, 46]
  • Social media and e-commerce are blurring: 70% of US users indulge in social networks \(\to\) 24% increase in “social shopping” earnings in 2017
  • The Challenge: Managing the complexity of simultaneous social and commercial activities

The Core Problem

Existing research lacks a detailed mapping of the “pathways” users take between social bonding and final transactions.

Research Question: How do users’ social and commercial activities affect user engagement (incidence and time spent) on a social commerce platform?

Conceptual Framework: Four Stages

Operationalized using Salience Theory across the platform lifecycle:

Social Identity & Interaction

  1. Social Identification — Viewing profiles, joining groups, establishing “who” belongs
  2. Social Interaction — Voting, liking, commenting, and conversing with other users

Shopping & Transaction

  1. Social Shopping — Browsing the product feed, viewing deals, and social-curated reviews
  2. Transaction — The final checkout and payment stage

The Transition: Understanding the “cost” (distance) and “importance” (rank) of these stages helps optimize platform design.

Methodology

Multi-disciplinary Approach

  • Computer Science Algorithms:
    • PageRank — Calculates Action Rank (importance of a specific node/action in a session)
    • Dijkstra’s Algorithm — Calculates Shortest Distance (the cost or number of clicks between actions)
  • Data: Real-world navigational clickstream data from a parental social commerce platform

Econometric Modelling

  • Multivariate Type-2 Tobit Model
  • Hierarchical Bayesian Methods (MCMC)
  • Models two dimensions of engagement:
    1. Incidence (The decision to visit/not visit a stage)
    2. Time Spent (The semi-latent variable of duration in each stage)

Key Findings

Rank and Distance Matter

  • Rank (+): Higher rank of social identifying/interaction states generally increases engagement in subsequent states.
  • Distance (-): Greater click distance between stages significantly reduces the likelihood of visit and time spent.
  • Asymmetry: Social shopping activities before transaction act as a serious distraction that can reduce social interaction engagement.

Engagement Context

  • Weekends/Evenings: Users visit social/interaction states more frequently but are less likely to enter the transaction state.
  • The “Shopping Barrier”: After commercial (shopping/transaction) activities, users are much less likely to return to social or entertaining activities in the same session.
  • User Permission: Granting rating/follower permissions significantly increases time spent across all stages.

Implications for Design

Theoretical Contributions

  • Operationalizes user engagement as a multi-stage, dynamic process rather than a static outcome
  • Integrates graph theory (ranking/distance) with established consumer behavior models (incidence/time)

Managerial Recommendations

  • Path Optimization: Minimize click distance from social interaction to shopping to prevent drop-offs
  • Strategic Ranking: High-rank interaction cues (like voting/likes) should be visually adjacent to products to bridge the social-commercial gap
  • Timing: Focus “Social Shopping” promotions early in the session; “Transaction” focus should avoid distraction from social interaction mid-funnel