“Automated attendance alerts” sounds like one feature. It is three, and conflating them is why so many schools end up either spamming parents daily or telling them nothing until the shortage notice arrives in March.
The three problems have different shapes, different urgencies, and different right answers.
Three Different Absence Problems, Three Packs
The pattern nobody noticed. A child has missed three days running. Somebody should know this week, not next month.
The cumulative shortfall. A child’s attendance for the month came in under the school’s threshold. This is a compliance and progression matter, and it needs to reach the family formally.
The register that was never marked. No amount of alerting works if the underlying data does not exist. In most schools this is the largest of the three problems and the one least often addressed by software.
Each gets its own pack, its own schedule, and its own thresholds.
Same-Day Absence and the Three-Day Streak
The absence streak pack watches for a student absent across a configurable number of consecutive school days — three by default. When the streak completes, the guardian receives a WhatsApp message, and the principal receives an aggregate summary of all streaks detected that day at a configurable hour, defaulting to 10:00.
Two settings decide whether this feels helpful or intrusive.
School days only, on by default. Absences are counted against days the school was actually open. Without this, a Friday absence followed by a weekend and a Monday absence reads as four days, and a school holiday can manufacture a streak out of a single missed day.
Immediate guardian notification, also on by default. When a streak is detected, the guardian message goes out that day rather than being batched. Schools that would rather have a staff member call before any automated message arrives can turn this off and keep only the principal summary.
Note what this pack deliberately is not: a same-day, every-day absence notification. The threshold exists because a message on every single absence trains parents to ignore the channel within a fortnight, and the messages that matter — the third day, the tenth day — arrive into an inbox that has already been tuned out.
In agent mode, the pack enriches each streak with the student’s rolling thirty-day attendance rate, their year-to-date rate, and how many prior streaks they have had over the last quarter. A structured call then classifies the streak — chronic, first-time, end-of-week pattern, or likely medical — and writes a short note to the class teacher alongside the guardian message. The classification changes who gets involved, not whether the parent is contacted.
The Monthly 75 Percent Warning Letter
The second pack runs on a fixed day of the month, defaulting to the 1st at 10:00, and evaluates the previous month’s attendance percentage for every student.
Two thresholds, not one:
- The warning threshold, 75 percent by default. Below this, the guardian receives a warning message naming the student, the actual percentage, and the required percentage.
- The at-risk threshold, 50 percent by default. Below this, the case is escalated in severity — the language shifts from a shortfall warning towards a serious-consequences warning.
Having two thresholds rather than one is what prevents the message being identical for a child at 74 percent and a child at 31 percent. Those are entirely different conversations and a single template flattens them.
In agent mode the differentiation goes further, producing first-warning, second-warning and at-risk variants based on where the student sits against your school’s policy, and raising a note to the class teacher with each.
The 75 percent default is not arbitrary — it reflects the attendance requirement most Indian boards apply for examination eligibility — but it is a setting, and schools with their own policy should set their own number rather than inherit ours.
Nudging the Teacher Before Nudging the Parent
This is the pack that fixes the upstream problem, and in most deployments it is the one that changes the numbers most.
It runs on a short cycle through the day. For each timetable entry scheduled today, roughly thirty minutes after the period ends, it checks whether any attendance record exists for that class and date. If none does, the teacher named on that timetable entry receives a push notification on their phone.
Three details make it tolerable rather than nagging. It fires in a narrow window after the period rather than repeatedly. It holds a deduplication key scoped to that specific period on that specific day, so the same class cannot generate two nudges. And it goes to the teacher’s phone as a push notification rather than to a school-wide channel — this is a private reminder, not a public flag.
The daily attendance summary that reaches the principal at 17:00 draws on registers that this pack has been quietly making sure exist.
Language, Channel, and Quiet Hours
Channel. Guardian alerts go out as WhatsApp template messages. Teacher nudges go out as push notifications to their device. Principal summaries and teacher notes are delivered in-app. These are not interchangeable — the channel is chosen per recipient type by the pack, and it reflects who is actually looking at what.
Language. The staff interface ships four language bundles: English, Hindi, Kannada and Tamil. Parent-facing message content is not bound by that list. In practice, what a parent can receive on WhatsApp is governed by which language variants of a template have been approved for your business account, which is a Meta process rather than a setting in the product.
Quiet hours. Attendance packs schedule their sends at working hours by design — 10:00 for the streak summary and the monthly warnings, 17:00 for the daily wrap. Separately, broadcasts on intrusive channels including WhatsApp, SMS and push are held during a school-default quiet window of 21:00 to 07:00 IST, with emergency-category messages exempt from it.
Consent and What Never Gets Sent
Attendance data is personal data about a child, and two boundaries apply.
Messages go to the guardian on record. If a family has no reachable contact number, no message is sent and no error is manufactured — the absence of a contact route is itself something the dropout risk scoring treats as a signal worth flagging to staff.
Separately, where attendance is captured through biometric or face-recognition methods, consent status is checked as a condition of access rather than as a one-time enrolment formality. An automation cannot route around a missing consent by asking for the data through a different door.
For the deduplication and circuit-breaker mechanics all these packs inherit, see school autopilot: operations that run without you.