Running a cohort so it behaves like a group
A cohort is not a list of people who bought the same thing at the same time. Making it a group is deliberate work.
Cohort-based courses are sold on the strength of the group: peers, discussion, shared momentum. In practice most cohorts are a set of individuals who never speak, in a chat channel that contains three introductions and then silence.
The group effect is real and it does not occur spontaneously. It is produced by a small number of specific design decisions.
A useful workplace parallel is attendance measurement: employee attendance tracking software shows how presence can be recorded separately from outcomes.
Size it for participation
Very small groups struggle when two people are absent. Large ones stop being groups: past a certain size nobody feels responsible for the silence and posting starts to feel like broadcasting.
Somewhere between eight and twenty is where most cohorts behave like groups. Larger cohorts can work if they are subdivided into stable smaller units that persist for the whole course — the persistence is what matters, because a group that is reshuffled weekly never forms.
For wider context on technology and learning, UNESCO publishes public resources on digital education.
Give the group something to do together
Discussion prompts produce discussion that reads like homework. What produces actual interaction is a task that requires other people: reviewing each other's work against a rubric, comparing different solutions to the same exercise, working in pairs on something neither can finish alone.
'Introduce yourself in the channel' produces a wall of introductions and nothing else. A task that cannot be completed alone produces conversation.
Establish the rhythm in week one and keep it
Cohorts run on rhythm: the same days, the same shape of week, the same points at which things are due and things happen.
Once established, that rhythm carries a lot of the accountability without anyone enforcing it. A cohort where the schedule shifts around, sessions move and deadlines slip loses the rhythm quickly, and after that every deadline requires individual chasing.
Notice people, visibly
The single most effective retention intervention in a cohort is a personal message to someone who has gone quiet. Not automated, not a group reminder — a short note that makes it clear a specific person noticed.
It works because most silent learners have not decided to quit; they have fallen behind and feel awkward. An acknowledgement that falling behind is normal, with a concrete suggestion about how to rejoin, converts a substantial share of them.
Decide how strict deadlines are, then be consistent
Deadlines that are enforced provide structure. Deadlines that are never enforced provide none. Deadlines that are enforced inconsistently produce resentment and are the worst of the three.
Either is defensible. A late policy stated at the start — a grace period, or a rule that work is reviewed only if submitted on time — is what makes it work. The one to avoid is a firm-sounding deadline with quiet exceptions for whoever asks.
Close the cohort properly
Courses often end by simply stopping: the last module is released, the channel goes quiet, and the group dissolves without acknowledgement.
A final session, a showcase of what people built, a clear statement that this is the end and what comes next — these cost one hour and they change what the experience is remembered as. They are also where testimonials and referrals come from, which is the commercial argument for something that is mostly worth doing anyway.