Time it once.
The number is the diagnosis.
Before changing anything, record one session or ask a colleague to sit in with a stopwatch, and measure the proportion of the hour in which a student — any student — is speaking. Most teachers estimate half and discover a quarter. It is not a comment on their teaching; explaining is the part of the job that feels most like the job, and a screen full of quiet faces pulls a teacher towards filling the silence.
Then do the arithmetic that follows from it, because it is stark. In a class of eight, a whole-class discussion gives each student, at absolute best, an eighth of whatever time is not the teacher. If the teacher takes forty of sixty minutes, the remaining twenty divided eight ways is two and a half minutes of speech per student per class. In pairs, the same twenty minutes gives each student ten. That is a fourfold difference produced by one structural decision, and it is the difference between a course where people improve and one where they attend.
This is not a claim from a study; it is division. But it explains why the schools that did well online are the ones that pushed production into small groups, and why the schools that struggled are often the ones with the most experienced, most articulate teachers.
A task with an outcome,
not a gap in the lesson.
Breakout rooms fail for one reason above all others: students are sent into them without a specific task, a time limit, or anything they have to come back with. Four people in a room with "discuss this" will exchange two sentences and wait. The fix is mechanical. Give a task that has an output — three differences, a decision, a ranked list, a question for another group. Say how long. Say what they will be asked for when they return. Then actually ask for it, because the first time a class reports back and nobody is asked, the rooms become social for the rest of the term.
Two more habits. Visit the rooms, briefly, rather than waiting in the main call — presence changes what happens in them, and it is also the only way to collect the errors you will correct at the end. And keep groups stable for a few weeks rather than reshuffling every session; a pair that has worked together three times talks considerably more than two strangers, which matters more online than in a room where they at least share a space.
On SprintUp the mechanics support this shape. You build a group set in advance and apply it to a live class, and the groups are copied at the moment the activity starts, so anything you improvise during the call does not silently rewrite the set you saved. Each group gets its own room, students move themselves into theirs rather than waiting to be assigned, and only one group activity can be running in a call at a time — enforced rather than left to memory, which removes the commonest source of chaos. The rooms close when you end the activity or when the meeting ends.
Cameras are
the wrong question.
Schools spend a lot of energy on whether to require cameras, and the energy is misplaced because the camera is a proxy for the thing actually wanted, which is participation. A mandatory camera policy produces students who are visibly present and mentally elsewhere, plus a subset who stop attending rather than show a bedroom they are embarrassed by or a family situation they would rather not broadcast. You have then traded real attendance for the appearance of it.
Design for participation instead and the camera question mostly dissolves. If every student speaks in the first five minutes, if the bulk of the hour is in pairs, and if groups report back, then a student who is not engaging is obvious regardless of what is on screen — and a student who is engaging is fine whether or not their camera is on. Where a camera genuinely helps is in small groups, and asking for it there, as a request with a reason, is both more effective and easier to accept than a blanket rule.
Five segments,
and where each one goes wrong.
The first row is the cheapest intervention in this article. If every student says something in the first five minutes — an answer, a reaction, one sentence — the session is established as one in which they speak. If the first twenty minutes are the teacher, the class has been told what kind of hour this is and will behave accordingly, and getting it back later is much harder than starting correctly.
The third row is where the hour is won or lost, and the discipline is to protect it when time is short. The instinct under pressure is to cut the group work and keep the explanation, because the explanation feels like the content. It is exactly backwards: the explanation can be sent as a note, and the production cannot.
Same day,
or not at all.
Online, a student who has decided to stop usually does not say so. They miss one, then two, then quietly become someone who used to attend. The single habit that changes this is contact on the day of the first missed session — short, friendly and specific, from the teacher rather than the office. It works because it arrives before the decision has hardened, and because it tells the student they were noticed, which is precisely what a building would have told them.
The platform can make the noticing reliable rather than dependent on someone remembering. In a live class SprintUp knows how long each student was connected and uses it to suggest a register entry against your own policy — the defaults are that at least 75% of the call counts as present and joining more than ten minutes after the start counts as late. It only pre-fills the register. The teacher confirms or overrides, because a connection that dropped for twenty minutes is not an absence, and marking it as one would be wrong about something a family will care about.