Amherst Student: Digital Transformation

Redesigning a 150-year-old college newspaper's digital presence, achieving 733% increase in readership through modern tech stack and machine learning

Role Digital Director
Year
Tech Stack
Next.jsReactPythonMachine LearningDrupal
Impact

733% increase in readership (1,200 to 10,000 monthly users), 85-90% newsletter read rates

Context

The Amherst Student, founded in 1868, is one of the nation’s oldest college newspapers. As Digital Director from 2018-2021, I led the complete digital transformation of the publication, growing readership from 1,200 to 10,000 monthly users while directly managing a mixed-discipline team of six - CS students learning web development, designers, and writers - focused on digital strategy.

Timeline: June 2018 - June 2021 Team: Six-person mixed-discipline team (developers, designers, writers) Scope: Full platform migration, ML-powered archival system, newsletter strategy, team training

Challenge

How do you modernize a 150-year-old institution’s digital presence while making decades of archival content discoverable and maintaining editorial excellence?

Key Problems

  1. Legacy Technology Stack Drupal-based CMS was difficult to maintain and limited our ability to ship new features quickly

  2. Archival Content Inaccessible 150 years of journalism was trapped in legacy systems with no effective way for readers to discover historical articles

  3. Declining Digital Engagement Traditional college newspaper model wasn’t resonating with modern student readership patterns

  4. Traffic Spikes and Reliability When news broke - like Jeff Sessions’ campus visit that drove 30,000 hits in a single day - the legacy infrastructure couldn’t handle the load

Solution

Complete platform modernization with focus on performance, discoverability, and editorial workflow:

Impact

733% increase in readership - Grew from 1,200 to 10,000 monthly users through improved UX and content strategy

85-90% newsletter read rates - Far exceeding industry averages through targeted content and engaged audience

Archival content discovery - NLP-powered topic classification made 150 years of journalism accessible and relevant to modern readers

Team leadership - Directly managed six-person mixed-discipline team, leading training and mentorship initiatives that helped non-technical members develop digital skills

Technical Details

The Next.js migration provided dramatic improvements in page load times and developer experience.

The recommendation system used NLP classification - specifically topic modeling and classification - rather than vector search. Articles were analyzed and categorized by topic, allowing the system to surface related archival content when readers engaged with current stories. When reading about campus housing policy in 2020, the system could surface relevant articles from the 1990s or earlier that covered similar themes.

This project demonstrated how modern web technology and NLP can breathe new life into legacy journalism institutions.