Higher Education
Of advising sessions now auto-booked
Retention improvement within two semesters
About
Several universities in a shared advising consortium wanted data-driven, always-on academic advising. Six full-time staff were booking 12,000 sessions a term, yet 24% of students failed to show and at-risk learners were slipping through the cracks. We built a risk-scoring and matching engine that pairs students with available advisors automatically and feeds outcomes into a shared analytics view.
Industry
Higher Education
Company size
1,000 – 5,000 employees
Founded
1971
The Company
A consortium of universities sharing an advising bottleneck
Several universities within a shared academic-advising consortium ran advising as a heavily manual, appointment-driven process: six full-time advising staff booking roughly 12,000 sessions a term across the participating campuses, working from static appointment calendars.
The volume of sessions meant advisors were largely reactive — meeting whoever booked a slot — rather than proactively identifying which students most needed an intervention before a small academic problem became a larger one.
The challenge
A quarter of sessions went unused while at-risk students slipped through
24% of booked sessions ended in a no-show, wasting advisor capacity that could have gone to a student who actually needed it, while students showing early signs of academic risk weren't being systematically identified or prioritised for outreach.
Booking was manual and calendar-based, with no mechanism connecting a student's actual risk profile to how urgently they should be matched with an advisor, or which advisor's expertise best fit their situation.
The Solution
Risk scoring and automated matching, not just a booking calendar
We built a BERT-based embedding model scoring student risk from academic and engagement signals, feeding a reinforcement-learning matcher that pairs at-risk students with available advisors based on both urgency and fit, then writes the booking directly to CalDAV calendars rather than requiring a manual scheduling step.
Engagement outcomes — attendance, follow-up actions, academic performance after the session — feed back into Looker dashboards, so the matching logic and the advising team both improve over time based on what actually worked.
The Results
92% auto-booked, retention up 6%
92% of advising sessions are now auto-booked by the matching engine, and retention improved 6% within two semesters as at-risk students were identified and matched proactively rather than waiting for a self-initiated booking. No-show rates fell sharply as sessions were matched with urgency and relevance in mind rather than simple calendar availability.
Advising staff shifted from spending most of their time on scheduling logistics to spending it on the substantive conversations the matching engine now ensures happen with the students who need them most.
KEEP READING
Explore more studies on AI, analytics, and enterprise intelligence.
