Student Success Early Warning Platform
Case Study

The challenge

An EdTech provider with 25,000 enrolled learners was experiencing a 34% dropout rate — significantly above industry benchmarks. Their student support team was reactive, only finding out about at-risk students after they had already disengaged.

What we built

• Multi-signal ML model combining login frequency, assignment completion rates, video engagement, quiz scores, and communication patterns
• Early warning flags generated 4-6 weeks before traditional lagging indicators (missed deadlines, grade drops) appear
• Automated intervention task creation for advisors with suggested outreach messaging based on individual risk factors
• Cohort health dashboards for programme managers showing population-level risk distribution by module and cohort
• A/B testing framework to measure intervention effectiveness and optimise outreach approaches

The results

The solution delivered measurable results within weeks of go-live. The client achieved 28% Dropout reduction and 4-6wk Early warning lead time, with full ROI achieved within 4 months Payback period.