Scaling Content Operations for a Global Learning Platform: A Content Architecture for edX
One of the world’s largest digital learning catalogs scaled past 4,000 offerings on a content architecture sized for a fraction of that. We rebuilt it as growth infrastructure — roughly 1,000 pages launched without adding editorial or engineering headcount.
The work
The Challenge: The Architecture Became the Limit on Growth
One of the world’s largest digital learning catalogs passed 4,000 offerings on a content architecture sized for a fraction of that — so every landing page, program launch and campaign variant routed through engineering, and the limit on growth was the platform’s architecture rather than the catalog or the brand.
The Strategy: Content by Domain, Pages by Composition
Rebuilt the content architecture as growth infrastructure, so marketing and platform could move at different speeds without stepping on each other.
- Content by domain, not by system. Course data stayed in the authoritative systems that own it; marketing and narrative content moved into Contentful as an orchestration layer.
- Pages composed, not built. Live course metadata, modular editorial content and design-system components assemble into a page.
- Governance built into the components. Brand and design rules are structural, instead of being enforced by reviewers after the fact.
The Outcomes: A Thousand Pages, No Added Headcount
Editorial capacity that scales with intent rather than with engineering.
- Growth | ~1,000 pages launched | no added editorial or engineering headcount
- Intelligence | 4,000+ offerings under one content model | SEO systemic across thousands of URLs
The detail
Why it mattered
- Discovery does not survive the catalog growing. A few hundred offerings can be browsed and searched. Past 4,000, across career switchers, working professionals, degree seekers and enterprise learners, they cannot. Organic and paid campaigns were sending traffic into a discovery experience that could not complete the match.
- Governance was maintained by making change slow. Brand consistency, compliance and design-system rules survived because every editorial movement routed through an engineering release. That is governance by friction, and it costs about as much as it protects.
- Editorial was coupled to the platform. Course data lived in edX’s authoritative internal systems, as it should. Marketing narrative, program pages and campaign surfaces were bound to those same systems, so any team that wanted to move faster was blocked by the platform’s architecture.
How it was built
- Content by domain, not by system. Course data stayed where it is authoritative. Marketing and narrative content moved into Contentful as an orchestration layer, which is the decision the rest of the engagement rests on.
- Pages assembled from three inputs. Structured course metadata from the internal systems, modular editorial content from Contentful, and reusable components governed by the design system. A page is composed rather than built.
- Content models sized to editorial reality. Modeled on how editors actually work, rather than on the platform artifacts they had previously been editing around.
- Performance as a system property. Edge image optimization across high-traffic pages, so faster global loads and better Core Web Vitals are a property of the architecture rather than a per-page concern.
What changed operationally
- Marketing regained velocity. Editors launch and iterate pages without developer support. New programs move through editorial rather than through engineering release cycles, and campaign pages adapt per audience and channel without bespoke builds.
- SEO became systemic. Structured content across thousands of URLs is indexable by design. Program pages can be optimized without breaking templates, and internal linking and taxonomy stay consistent because the architecture enforces them.
- Governance scaled with the content. Design-system enforcement is structural rather than review-based, and brand and compliance requirements are properties of the components themselves — not because a reviewer catches violations after the fact.
Content operations at scale questions
When does content architecture become a growth problem?
When the catalog outgrows the system that presents it. A few hundred offerings can be browsed and searched; past 4,000 they cannot, and every new program launch, campaign page and SEO test at edX routed through an engineering release. At that point the limit on growth is the platform’s architecture rather than the catalog or the brand.
Should marketing content live in the same system as product data?
No — each should live where it is authoritative. edX’s course data stayed in the internal systems of record; marketing and narrative content moved into Contentful as an orchestration layer. Pages assemble from both, so the marketing team moves at its own speed without duplicating the data or touching the platform.
How does a CMS improve SEO at scale?
By making it systemic instead of manual. Structured content across thousands of URLs is indexable by design, program pages can be optimized without breaking templates, and internal linking and taxonomy stay consistent because the architecture enforces them.
How do you keep brand governance without slowing editors down?
Move the enforcement into the components. On the old edX platform, governance was maintained by routing every change through engineering review — governance by friction. In the rebuilt architecture, design-system rules, brand standards and compliance requirements are properties of the components a page is composed from, so editors move fast inside them.
What are the benefits of a content re-architecture?
For the edX rebuild by Pare & Co: roughly 1,000 pages launched with no added editorial or engineering headcount, program launches that move through editorial instead of engineering release cycles, and SEO that is consistent across thousands of URLs by design.
Client leadership
Doug Sisko
Doug Sisko’s specialty is growth and conversion strategy. He has led content operations work where publishing volume was the growth constraint.