Most schools look at one attendance number: the overall rate, this month, across everybody. It moves between 88% and 93% and tells you nothing you can do anything about.
The same data cut four other ways is one of the most actionable things you own.
By class, over time
This is the cut that matters most. A single class dropping from 92% to 78% over three weeks is a specific, fixable signal — a teacher problem, a clash with school exams, or a group that has stopped enjoying it.
The overall figure will not move enough to notice, because your big classes drown it out. That is why the overall figure feels useless: it is.
By student, as a streak
Not "78% this term" but "missed the last two". A student at 78% who attends every week now is fine. A student at 91% who has missed the last two sessions is leaving — see the students who quietly leave.
Percentages describe the past. Streaks predict.
By day and time slot
Your Tuesday 5pm may run at 95% while Saturday 9am sits at 74%. That is not a student-quality difference, it is a scheduling insight: the Saturday slot is fighting family time, and no amount of teaching fixes it.
Two responses, both legitimate. Move the class, or accept the lower rate and stop reading it as a problem.
By teacher, carefully
Useful, and the easiest to misuse. Attendance varies with level, age group and time slot far more than with teaching quality — a Saturday morning kids' class will always trail a weekday exam-prep group.
Compare like with like or not at all. And treat it as a question to ask, never as a performance score. A teacher who believes attendance is their appraisal will start marking generously, and then you have lost the data entirely.
The moment a number decides someone's pay, it stops being a measurement and becomes a negotiation.
The number nobody tracks
Sessions with no attendance marked at all. Every one is a hole in every figure above, and in the parent portal.
Track that as its own count and chase it weekly. Data quality is not glamorous, and every other number here depends on it.