The Real Problem
James is the academic manager at an English language school in Auckland CBD. 120 students across six levels, four classrooms, eight teachers. Every morning, teachers mark a paper roll. At the end of the week, the admin assistant enters attendance into SELMA. On Monday, James reviews the numbers.
Three weeks ago, a Japanese student named Yuki stopped coming to class. She attended sporadically for the first week, one day on, two days off, then stopped completely. By the time James noticed the pattern in the weekly data, Yuki had been absent for five consecutive days. He emailed her. No response. He called. Voicemail. He contacted her homestay family, who said she'd been staying in her room, not eating much.
It turned out Yuki was struggling with loneliness and culture shock. She hadn't told anyone because she didn't want to be a burden. If someone had noticed on day two, they could have connected her with the school's pastoral care coordinator, arranged a conversation with a Japanese-speaking counsellor, or simply invited her to a student activity. Instead, she sat alone in her room for a week before anyone knew.
This isn't rare. Industry data shows that English language learner dropout rates run 25-27.5% across the sector. The longer a student remains disengaged, the greater the dropout risk. And in New Zealand, student attendance isn't just an academic concern. It's a visa compliance requirement.
Immigration New Zealand requires language schools to track and report student attendance. If a student on a student visa falls below the attendance threshold, the school is obligated to report this. Failure to track and report accurately puts the school's NZQA registration at risk. Schools have been deregistered for compliance failures.
Most schools track attendance. Very few analyse it in real time. The data sits in a spreadsheet or student management system, reviewed weekly or monthly, always after the fact. By the time someone notices a pattern, the student is already gone.
Why Existing Tools Don't Solve This
SELMA SIS is excellent at recording attendance and generating compliance reports. It tracks student hours, calculates percentages, and produces the data NZQA requires. But SELMA is a record-keeping system, not an early warning system. It tells you what happened. It doesn't tell you what's about to happen.
Class by Infospeed similarly tracks attendance and can flag students below a threshold. But the flag triggers after the student has already fallen below the required percentage, not when the pattern is emerging.
Moodle and Canvas track engagement with online learning materials, but most NZ language schools run primarily face-to-face programmes. LMS engagement data doesn't capture the student who shows up to class but has checked out emotionally, or the one who missed two days and is about to miss three more.
Manual monitoring by teachers is unreliable at scale. A teacher with 15 students in the morning and a different 15 in the afternoon can notice that one student seems quiet, but they can't systematically track patterns across all students over time. And they certainly can't cross-reference attendance with assessment performance, homework submission, and social activity participation.
No existing tool in the NZ language school market combines attendance data with behavioural patterns to predict which students are at risk of disengaging before they actually disappear.
How AI Solves This
The school connects its attendance data (from SELMA, Class, or even a simple daily spreadsheet) to an OpenClaw AI assistant. The AI analyses patterns daily and generates alerts when a student's behaviour suggests they're at risk.
Tuesday morning, 9:30am. The AI sends James a notification:
At-Risk Student Alert: Yuki Tanaka (Level 4, Japanese)
Attendance trend: 100% (Week 1-3) > 80% (Week 4) > 60% (Week 5, current) Pattern: Missed Monday and Tuesday this week. Last week missed Thursday and Friday. Assessment: Last two weekly tests scored 45% and 38% (previous average: 72%) Homework: Last 3 submissions missing Risk level: HIGH
Suggested actions:
- Pastoral care check-in today (Japanese-speaking support available)
- Academic review with Level 4 teacher
- Attendance currently at 78%. NZQA threshold is 80%. Two more absences this week will trigger reporting requirement
James sees this on Tuesday morning, not the following Monday. He walks to the pastoral care coordinator's office. By noon, someone is calling Yuki. By Wednesday, she's meeting with a Japanese-speaking counsellor. By Thursday, she's back in class with a modified support plan.
What the AI Monitors
The system tracks multiple signals across each student:
- Attendance patterns including day-of-week trends, late arrivals, and early departures
- Academic performance from weekly tests, level assessments, and teacher evaluations
- Homework and assignment submission rates and timeliness
- Social engagement indicators such as activity participation and excursion attendance
- Visa compliance thresholds calculated in real time against NZQA requirements
The AI doesn't just flag students who have already fallen below thresholds. It identifies declining patterns and predicts which students are trending toward disengagement.
Automated Compliance Reporting
Beyond early warning, the AI also generates:
- Weekly attendance summaries by level, nationality, and visa type
- NZQA-formatted attendance reports ready for submission
- Visa compliance alerts when a student approaches the reporting threshold
- Annual return data pre-formatted for NZQA submission deadlines
What used to take the admin team a full day every Friday now takes 15 minutes of review.
How We Set This Up
None of this works if the AI is just a standalone analytics tool with no connection to your actual student data. That's why BestAI builds a custom integration program that bridges your AI assistant with the systems you already use.
For this kind of setup, that means:
- Connecting the AI to your student management system (SELMA, Class, or equivalent) for real-time attendance data
- Setting up daily analysis routines that run automatically each morning
- Configuring alert thresholds based on your school's policies and NZQA requirements
- Building a dashboard for academic managers and pastoral care staff
- Integrating email and SMS notifications for urgent alerts
Here's our process:
- We map your data sources - We identify where your attendance, assessment, and student information currently lives and how it flows.
- We build the connections - Our developers write a custom program (an API connector) that lets the AI access your student data securely, in real time.
- We configure your thresholds - Every school has different policies. We set up alerts that match your specific attendance requirements, visa compliance thresholds, and academic benchmarks.
- We test and refine - We run the system alongside your existing processes for two weeks, comparing AI alerts against manual observations, before fully deploying.
You don't need to be technical. We handle all the development. You just tell us what patterns you want to catch, and we build the system to catch them.
The Result
- At-risk students identified 3-5 days earlier than manual monitoring
- Pastoral care interventions happen while they can still make a difference, not after the student has already withdrawn
- NZQA compliance reports generated automatically with accurate, real-time data
- Attendance tracking takes 15 minutes per week instead of a full day
- Visa compliance alerts prevent reporting failures that could jeopardise NZQA registration
- Student retention improves measurably because early intervention actually works when it's early
For a school where each student pays NZ$400 per week, every student retained for an additional 4 weeks through early intervention represents NZ$1,600 in revenue that would otherwise walk out the door. Retain 5 students per term, and that's NZ$8,000 in preserved revenue, plus a student who got the support they needed.
What AI Can't Do Here
- AI won't counsel students. It identifies who needs help. Your pastoral care team provides the actual support
- AI won't make visa decisions or report students to Immigration NZ. It flags when thresholds are approaching. Your compliance team decides what action to take
- AI won't override teacher assessments or academic judgements
- AI won't contact students directly. Outreach is always initiated by staff who know the student
- AI won't diagnose the reason for disengagement. It spots the pattern. A human conversation reveals the cause
Who This Is For
- Language schools with 50+ students where manual monitoring can't catch every pattern
- Schools that have had compliance issues with NZQA attendance reporting
- Academic managers who review attendance data weekly and wish they'd seen problems sooner
- Schools with high proportions of students on student visas (where attendance tracking is a legal obligation)
- Any school that has ever lost a student who could have been saved with a conversation three days earlier