A CBT group runs to a protocol. Eight to sixteen weeks, a defined curriculum, an agenda each session, homework, and measures taken at the start and the end so you can both see whether it worked.
It’s the closest thing in mental health to a course with a syllabus, and that’s the point rather than a limitation.
What the evidence says
Group CBT performs comparably to individual CBT across anxiety and depression in most head-to-head comparisons — which, given it costs a third as much and treats six to ten people at once, is a striking finding that has never really made it into public awareness.
For social anxiety the case is stronger still, and not for economic reasons:the group supplies the feared situation. Individual therapy asks you to describe being scrutinised by a room of people. A group is a room of people.
What a session contains
An agenda, agreed at the start. Homework review. New material taught with examples. Practice in the room — and this is the part that distinguishes a good CBT group from a lecture: behavioural experiments, role-play, exposure exercises done together.
Then homework set for the week, because in CBT the homework is the treatment and the sessions are where you learn how and troubleshoot.
The common kinds
Social anxiety groups — the strongest indication, usually twelve to sixteen weeks, heavy on in-session experiments.
Anxiety and panic groups — including interoceptive work, which is easier in a group than people expect because watching someone else do it first lowers the bar.
Depression groups — behavioural activation with people in it, which addresses the withdrawal directly rather than discussing it.
CBT-I for insomnia — frequently delivered in groups and highly effective.
Mindfulness-based cognitive therapy (MBCT) — group-delivered by design, with good evidence for preventing depressive relapse.
Who it suits
People with a specific, nameable problem who want a defined end. People who would find an open-ended process group intolerable. People who like homework, or at least tolerate it.
And people who want to know whether it worked — the measures are a genuine advantage of the format.
Who it doesn’t
If the difficulty is how you’re with people rather than a specific symptom, a protocol group will not touch it, because it’s deliberately not looking at what happens between members.
And if what you’ve is OCD, the group needs to be an ERP group specifically —general CBT applied to obsessions is a compulsion, and that’s as true in a group as anywhere.