CMS Architecture & Authoring Experience

 

Rebuilding Penn State’s CMS Architecture & Authoring Experience

Before redesigning Penn State’s CMS strategy, every site maintained its own content structure. Models were duplicated. Naming conventions were inconsistent. Editors had drastically different workflows depending on which domain they were working in. There was no unified preview workflow, no way to share components or schemas across spaces, and no tooling to govern changes across the organization.

This fragmentation created operational bottlenecks and development overhead. It also made content governance, brand consistency, and long-term scalability nearly impossible.

My work focused on designing a modern CMS architecture—built on Contentful—that supported modular content models, predictable authoring experiences, and a suite of enterprise-grade internal tools.

Multi-Tenant Content Machine

The foundation of the system was a modular content model architecture. Instead of building large, rigid content types, I broke them down into reusable, composable modules that could be shared across spaces and reused in multiple site contexts.

This provided several immediate benefits:

  • Editors gained predictable patterns for structuring pages.

  • Developers could reference a single schema instead of rebuilding types per site.

  • Updates to modules propagated cleanly across the ecosystem.

  • Each tenant site could adopt only the modules it needed, reducing clutter in the UI.

This modular strategy became the backbone of how pages, components, and structured content are represented throughout Penn State’s digital properties.

Improving Preview Workflows & Author Experience

Before these improvements, faculty and marketing teams had no reliable way to preview their work—especially on dynamic pages driven by multiple entries and multi-level relations. I rebuilt the preview workflow end-to-end, including:

  • A real-time preview mode integrated with Next.js

  • The ability to test multiple states of a page (draft, staged, published)

  • Context-aware previews that reflect localized, per-space content

  • More intuitive status messaging and error handling for editors

These changes gave editors a confident, stable preview experience that matched what would ultimately appear on the live site.

Dashboard: Content Model Syncing Across Spaces

Penn State operates many Contentful spaces—Sandboxes, QA environments, staging spaces, production domains, and program-specific tenants. Each needed access to the same base models, but sync was previously manual and error-prone.

I built a Content Model Sync Dashboard that:

  • Compares schema differences between any two spaces

  • Highlights missing or outdated models

  • Allows one-click synchronization of content types and fields

  • Reduces schema drift across environments

This eliminated a major operational burden and standardized our CMS architecture.

Exporting Content Across Spaces

I built a second dashboard for cross-space content export, enabling teams to:

  • Select entries, models, or groups of content

  • Export them from one space

  • Re-import them cleanly into another

This is used for onboarding new sites, migrating program content, and seeding QA environments with real data.

Previously, migrations required engineering assistance. Now, non-technical team members can move structured content without developer intervention.

A/B Testing for CMS-Driven Variants

Traditional A/B testing tools don’t integrate naturally with CMS-driven sites, especially when many page variations contain structured content or deep references.

I built a Contentful A/B Testing Dashboard that allows:

  • Creating test variants inside the CMS

  • Assigning audience weights

  • Previewing each variant

  • Deploying variant configurations through our API layer

This bridges the gap between editorial teams and experimentation workflows.

Content Quality Audit with LLM-Powered Clustering

Finally, I designed an internal Content Quality Audit Dashboard that uses LLMs to analyze, score, and cluster content across large sites.

It identifies:

  • Duplicate or low-value pages

  • Outdated or underperforming content

  • Pages with weak titles, poor metadata, or inconsistent tone

  • Content clusters that could be consolidated

The dashboard gives editors actionable insight into how to prune and refine their site at scale—something previously impossible without manual auditing.

The Result: A Modern, Scalable Content Architecture for Penn State

Collectively, this work transformed Penn State’s CMS environment from a collection of isolated, inconsistent spaces into a unified, modular, tool-rich ecosystem.

The outcome:

  • Modular content models standardized across sites

  • Reliable preview workflows that editors trust

  • Dashboards for governance: model syncing, content export, A/B testing, auditing

  • More efficient editorial workflows with reduced dependence on engineering

  • A scalable foundation supporting every program, college, and platform team

The new CMS architecture not only improved productivity—it created a long-term governance framework for how content is created, previewed, tested, migrated, and retired across one of the largest university networks in the country.

Previous
Previous

Myriad

Next
Next

Marketplace App