THE CHALLENGE

Professor Paul Resnick and the University of Michigan School of Information needed a research platform enabling academic study of social media behavior - but existing approaches were fundamentally limited. Social media companies conduct internal experiments on their own platforms, but no broad research infrastructure existed for studying user behavior across multiple platforms from a person-centric perspective.

The complexity

• Build cross-platform monitoring (browser extensions for Chrome, Firefox, Safari + mobile apps for iOS and Android)
• Track user behavior across multiple social media sites simultaneously (Facebook, Twitter, Reddit, Instagram, etc.)
• Collect sensitive behavioral data (time spent, posting patterns, content exposure) with strict IRB compliance and privacy protections
• Enable intervention studies where researchers can present surveys and experimental treatments
• Support longitudinal studies tracking behavior over months/years
• Process and analyze behavioral data at scale while ensuring participant anonymity
• Integrate with external research studies (e.g., Colorado adolescent substance use study linking social media to health behaviors)

The research question demanded person-centric data - understanding how individuals split time across platforms, how usage patterns change over time, and correlations with other behaviors - but existing tools only provided site-centric views. Building this infrastructure meant solving technical challenges (cross-platform monitoring, offline data collection, privacy-preserving analytics) alongside research methodology challenges (IRB compliance, informed consent, participant retention).

University of Michigan chose AnvilEight based on Python/Django expertise, proven mobile development capability (iOS + Android), and understanding of academic research requirements including ethics compliance and multi-year project timelines.

THE RESULTS

A 10+ year partnership delivering research infrastructure enabling groundbreaking social media behavioral studies, funded by Google Research, Facebook, and NIH.

10+ Years Partnership with U-Michigan

$500K+ Google Research Funding

Cross-Platform Browser + iOS + Android

Multiple Studies NIH, Social Media, Health Research

Research Infrastructure Delivered

• Browser extensions (Chrome, Firefox, Safari) monitoring social media usage across platforms
• Mobile apps (iOS and Android) tracking mobile social media behavior and location
• Backend research platform processing behavioral data and delivering interventions
• Survey delivery system presenting questionnaires on participants' devices
• Data analytics infrastructure enabling researchers to study cross-platform behavior patterns
• Privacy-compliant architecture with IRB approval for sensitive behavioral tracking

Research Enabled

• Google-funded MTogether platform studying social media usage patterns across platforms
• NIH substance use study ($272k sub-award) linking adolescent social media behavior to substance use
• CU Boulder collaboration tracking 400 adolescent twins over one year
• Annual State of Social Media reports presenting findings to corporate sponsors
• Academic publications advancing understanding of social media behavior and health outcomes

Technical Achievements

• 10+ years platform stability maintaining infrastructure across platform API changes
• Cross-platform consistency collecting comparable data from web (3 browsers) + mobile (iOS + Android)
• Privacy-preserving architecture enabling sensitive behavioral research with IRB approval
• Scalable data processing handling behavioral tracking from hundreds of research participants
• Longitudinal data quality maintaining participant engagement over months/years of study

Testimonial

"Andrii has pulled together teams for me on different projects over the past few years. They specialize in Python-based web development (Django, Google App Engine) and mobile apps (iOS and Android). I've learned a lot about good development practices from them, some of which I now try to teach to my students!"

Professor Paul Resnick - University of Michigan School of Information

TECHNICAL CHALLENGES SOLVED

Challenge 1: Cross-Platform Behavioral Monitoring

Problem: How do you collect comparable behavioral data across web (Chrome/Firefox/Safari), iOS, and Android when each platform has different APIs, permissions models, and technical constraints?

Solution: Built platform-specific implementations sharing common data models and backend APIs. Browser extensions used JavaScript with platform-specific permission handling; iOS used CoreLocation and background processing; Android used different approaches across API versions. Backend abstracted platform differences, providing researchers unified datasets regardless of participants' device choices.

Challenge 2: Privacy-Compliant Sensitive Data Collection

Problem: Tracking social media behavior involves sensitive information requiring university IRB approval. How do you collect research-necessary data while protecting participant privacy and meeting ethics requirements?

Solution: Architected privacy protections comprehensively: informed consent workflows explaining tracking scope, participant controls to pause/disable monitoring, data anonymization separating identity from behavior, secure transmission/storage with encryption, and IRB-approved data retention policies. Privacy protection enabled ethics approval, which enabled research - technical architecture determined research feasibility.

Challenge 3: Long-Term Platform Stability During API Evolution

Problem: Social media platforms constantly change APIs, authentication mechanisms, and data access policies. How do you maintain research infrastructure over years when underlying platforms shift?

Solution: Abstracted platform-specific code into adapters isolating API changes, implemented robust error handling recovering from authentication failures and API rate limits, maintained comprehensive logging identifying breaking changes quickly, and established update processes minimizing disruption to ongoing studies. Platform evolution couldn't break multi-year research - our architecture ensured continuity.

Challenge 4: Participant Retention Over Longitudinal Studies

Problem: Academic research requires tracking participants over months/years, but browser extension/app installations often get abandoned. How do you maintain participant engagement?

Solution: Minimized performance impact (lightweight monitoring not slowing browsers/devices), provided participant value (personalized usage reports showing their own patterns), implemented gentle reminders without nagging, and designed unobtrusive monitoring not interfering with normal usage. Technical quality enabled research methodology - participants stayed enrolled because technology worked well.

THE OUTCOME

Ten years supporting groundbreaking social media research at University of Michigan, delivering technical infrastructure enabling studies shaping academic understanding of online behavior, health outcomes, and platform effects.

For researchers: Platform enabling research questions impossible without person-centric cross-platform data collection. Published findings in top academic venues, secured continued funding (Google, Facebook, NIH), and established Michigan as leader in social media research methodology.

For participants: User-friendly tools providing personal insights (My Social Media Report) while contributing to academic research advancing societal understanding of social media's effects.

For social media ecosystem: Annual State of Social Media reports synthesizing research findings, shared with corporate sponsors (Google, Facebook) and academic community, advancing industry and scholarly understanding of platform usage patterns.

For AnvilEight: Longest academic partnership (10+ years) demonstrating capability supporting complex research infrastructure, maintaining platforms through technology evolution, and balancing technical sophistication with research methodology requirements and ethics compliance.