We see students leaving before they do.

Pakistani academies lose 15–25% of their students to dropout every year. For a 300-student school at Rs 5,000/month per student, that is Rs 1.8–3.75 million in annual revenue lost — not from students who chose a better school, but from students whose departure had early warning signs that nobody caught in time. Declining attendance. Slipping grades. Late fees three months in a row. A parent who stopped reading circulars. A student who used to borrow library books and suddenly stopped. These are all signals. Individually, they might be noise. Together, they predict a dropout 6 weeks in advance — if someone is watching.

The Dropout Prediction engine in Kliovo Edu watches all 42 data points simultaneously, for every student, every week. It does not require any input from teachers. It does not require the principal to review a spreadsheet. Every Sunday night at 11 PM, the engine runs a full recalibration: it re-scores every student across all data dimensions and updates the watchlist. Monday morning, the principal opens the dashboard and sees a color-coded list: green (on track), yellow (watch), orange (soft action needed), red (intervention required). The flags are never more than 7 days stale.

The design philosophy is privacy-first. Parents never see the 'at-risk' label — it is internal, admin-only. When ovo AI sends a soft outreach message on Orange or Red students (a gentle 'We noticed Hasan hasn't been in school for a few days — is everything okay?'), it does so without revealing the risk classification. The teacher sees a subtle indicator for their own students only — not a full risk report, not other teachers' students. The principal sees the full picture. This scoping is enforced at the database level, not just the UI.

Every flag is fully auditable. A principal can click on any at-risk student and see exactly which data points triggered the flag: 'Attendance: 62% (7 missed days this month). Fees: 22 days overdue. Parent: 0 circulars acknowledged in 3 weeks.' There is no black box. The system does not make decisions — it surfaces patterns. The principal and teachers make the intervention decisions. The engine's job is to make sure those decisions are made with enough time to act, not after the TC is already submitted.