Skip to content
CDO On-Demand
← All case studies

Unifying student data across 125,000 learners on Google Cloud

A remote learning institution serving more than 125,000 students generated rich data across student information, learning management, and curriculum systems, none of which spoke to each other. A governed BigQuery and Looker platform brought those streams together to support early intervention, faculty insight, and executive reporting.

Client
Remote learning institution
Engagement
Data platform design and delivery leadership

20 weeks

From project initiation to full production deployment

125,000+

Students covered across integrated systems

4 personas

Dashboard tiers from executive to student success advisor

Figures above were the targets this program was designed and approved against, rather than post-implementation measurements.

The outcome figures below were the targets the program was designed and approved against, rather than post-implementation measurements.

The situation

The institution had pioneered remote learning since 2010 and served more than 125,000 students through live classes, interactive sessions, and collaborative projects. That activity produced substantial and genuinely valuable data across three system families: student information systems holding enrollment, grades and records; learning management systems carrying engagement and submission behavior; and curriculum design systems supporting faculty content development.

The systems worked. What did not exist was any connection between them, which meant nobody could answer questions that spanned more than one, and the questions that mattered most about student success all spanned several.

The approach

A governed foundation on Google Cloud. Cloud Storage as the initial repository for raw and processed data, BigQuery as the analytical engine, Cloud Dataflow handling both real-time and batch processing, and Cloud Composer orchestrating the workflows and maintaining data quality on schedule.

Integration across the three system families. Secure connectors captured enrollment, grades, demographics, and progression from student information systems; participation, submission patterns, forum engagement, and resource access from learning management; and content performance and pathway effectiveness from curriculum systems.

Dashboards designed by role rather than by data source. Four tiers, each built around the decisions that role actually makes:

  • Executive leadership: enrollment trends, retention, financial performance
  • Academic operations: course performance, curriculum effectiveness, resource use
  • Faculty: engagement insight and assignment performance for their own students
  • Student success: early warning indicators and intervention triggers for advisors

The outcome

The platform reached full production twenty weeks from initiation, following a pilot with a limited user group and a phased departmental rollout. The program was designed and approved against targets of 35% improvement in retention through early intervention, 50% faster time to actionable reporting, and 25% improvement in resource allocation efficiency.

What made it work

Six weeks of discovery before any construction began. Stakeholder workshops with faculty, administrators, and IT established what each group actually needed to decide, which prevented the common failure of building dashboards that reflect the shape of the source systems rather than the shape of the decisions.

Why this is relevant to you

Business, technical, and program together.

The business lens

The institution cared about retention, student success, and whether faculty could see a struggling student early enough to intervene. Those outcomes drove the design, which is why the dashboard work was organized around what each role needed to decide rather than around what the source systems happened to contain.

The technical work

Student information, learning management, and curriculum design systems each hold a partial view of the same learner, with different identifiers and refresh characteristics. Building a coherent longitudinal record required careful entity resolution and a pipeline design that handled real-time and batch sources without producing two versions of the truth.

Program and organization

Twenty weeks from initiation to production across an institution serving 125,000 students required a phased go-live, structured user acceptance testing with representative groups, and role-based training running from IT administration through faculty to executive leadership.

Services

Data platform architectureDashboard designData governance

Stack

BigQueryLookerCloud DataflowCloud ComposerCloud Storage

Have a similar problem?

If that resembles the situation in your own organization, a short call is the quickest way to establish whether the same approach would apply to you.