Students who leave a college rarely decide suddenly. The measurable precursors — attendance falling against their own baseline, a missed internal assessment, accumulating backlogs, an overdue fee instalment — typically appear weeks before withdrawal and are already in the institution’s own records.
Signals identify who to talk to, not why. The reasons are financial, academic or personal, and only a conversation distinguishes them.
Most colleges discover a dropout when a student stops appearing, and record it as a statistic. By that point the decision was usually made weeks earlier, and often could have been changed.
The data needed to see it coming is already held. It is just held in four places and looked at by nobody together.
A 75% threshold flags students who have always sat near it and misses the more informative case: a student who attended everything for two semesters and dropped to 82% this month. The first group contains few dropouts; the second contains many.
Change against a student’s own pattern is the signal. Absolute thresholds are for eligibility, which is a different question.
An automated email to a student showing early signs of disengagement is worse than nothing. It confirms that the institution has noticed and does not care enough to send a human.
What works is a named mentor, with a small enough list to actually call, having a conversation in the week the signal appears. The system’s only job is to produce that list early and keep it short enough to be actioned.
Only the second is an academic problem. Treating all four as academic problems is why many retention initiatives fail.
Phrased the way they arrive, answered so each one stands on its own.
Predominantly financial pressure, accumulated academic backlog, programme mismatch and personal circumstances. Only one of those is academic, which is why interventions framed purely around academic support have limited effect.
Typically several weeks before withdrawal, sometimes a full semester, using attendance change against a personal baseline plus missed assessments. Early enough to matter, if someone acts on it.
No. Being told a system considers you likely to leave is more likely to produce the outcome than prevent it. The flag exists to prompt a mentor, not to inform the student.
Checked by the Implementations desk on 1 September 2026. Corrections go in the page, dated, rather than quietly.
Pricing is on the page, in the open, with the counting rule written next to it. Nobody has to book a call to find out what it costs.
See pricingForty minutes with an implementation engineer — your classes, your fee heads, your grading scheme, entered as we go. No slides, no sample school.