Methodology

How the early-warning model works

No black box. Faculty can see which signal moved a score and why a particular intervention was suggested.

Collect

Attendance registers, assessment results and learning-platform activity sync nightly from your student information system.

Signals are scaled per course and per cohort, so a hard module doesn't flag an entire class as failing.

A weighted model produces a 0–100 risk score plus a predicted end-of-term grade, refreshed every Monday.

Each flagged student is matched to an intervention type, an owner and a deadline based on which signal is failing.

Tuned against three prior cohorts of end-of-term outcomes.

What the platform suggests when a specific pattern appears.

Attendance falls below 70% or drops 15 points in 6 weeks
→ Attendance contract plus a mentor check-in within the week
Two consecutive assessments below 50%
→ Diagnostic test and a structured study plan for that subject
Three or more missed submissions in one course
→ Peer tutoring pairing with a teaching assistant
Engagement collapses while grades hold
→ Wellbeing conversation with student services before grades follow

Scores are advisory. They open a conversation; they never decide a grade or a progression outcome.
Only staff assigned to a student can see that student's profile and history.
Every flag records the signals behind it, so a student can be told exactly why they were contacted.