Week 8

Network cohesion and community - case discussion

Published

December 10, 2025

Prepare

Note📚

📖 Review Week 7 materials on cohesion, community detection, and attention-based segmentation

📖 Read the case study: “Fuller’s and the Problem of Market Segmentation in the Beer Market” (ba_case.pdf)

📌 Focus on Exhibits 2-4: contrasts between traditional vs. network-based segmentation, the two-mode user-style representation, and the three strategic options

📝 Jot down where BeerAdvocate attention data could reveal segments that Fuller’s current research misses

Participate

Note👥

🙋 Case Discussion: Fuller’s attention-based segmentation dilemma

Sarah Mitchell and James Chen must choose how to re-segment the beer market as craft, imports, and digital discovery squeeze Fuller’s “middle” position. The board needs evidence that attention networks from BeerAdvocate can guide portfolio and brand decisions.

Key Discussion Questions:

  • What signals about loyalty, exploration, and cross-over styles are invisible to traditional focus groups and surveys?
  • How would you build and validate a two-mode network (users ↔︎ beer styles) from BeerAdvocate, and what biases/coverage limits should Fuller’s anticipate?
  • Compare Options 1-3 (enhanced traditional research, network analysis pilot, integrated platform): which path fits Fuller’s risk tolerance, timeline, and capability needs?
  • How could attention communities translate into concrete moves (range rationalization, innovation briefs, pub assortment, messaging) for London Pride, ESB, and newer styles?
  • If you recommend a pilot, what success metrics and governance would make the board confident to scale?

Practice

Note💻

Hands-on Analysis: Attention-based segmentation

📊 BeerAdvocate Network Analysis

Explore how attention flows across beer styles and communities of enthusiasts:

  • Build the two-mode user-style network and project it to a user-only network
  • Run community detection to surface attention clusters and interpret “taste brokers”
  • Compare degree, betweenness, and clustering to spot core vs. peripheral styles and users
  • Experiment with different time windows or style groupings to test segmentation stability

Perform

Note📝

Case Prep (bring to class)

  • Choose your preferred option (1: enhanced traditional, 2: network pilot, 3: integrated platform) with 2-3 bullets of rationale
  • Outline the minimum viable pilot: data scope, metrics of success, and how insights would feed portfolio decisions
  • Note the biggest risks (selection bias, capability gaps, cost/pace) and how you would mitigate them

Ponder

Note🤔

Attention, community, and heritage

  • How does segmenting on shared attention (communities of discussion) differ from demographics or occasions, and when is each lens most actionable?
  • Where might “taste brokers” who bridge styles signal early shifts in demand, and how could Fuller’s engage them?
  • How can a heritage brand balance authenticity with experiments prompted by online community signals?
  • What safeguards are needed when blending public community data with proprietary sales and pub insights?

Further Reading:

  • Aral, S. (2020). The Hype Machine - attention dynamics and diffusion in online communities
  • Wasserman, S., & Faust, K. (1994). Social Network Analysis - foundational methods for two-mode networks and projections
  • Blei, D. M. (2012). “Probabilistic Topic Models” - alternative ways to cluster unstructured community text