The post-mortem after social events, and why it hurts
The replaying afterwards is where social anxiety does a lot of its damage. Here is what the post-mortem is actually doing, and how CBT stops it rewriting the evening.

The event is over. You got through it. And now, on the way home, or in bed, or at random moments for the next three days, you are back in it. Replaying the conversation, finding the moment you got it wrong. The pause that went on too long. The joke that did not land. The thing you said that, on the fifth replay, sounds worse every time.
Other people describe the same evening as perfectly ordinary. You do not believe them.
The post-mortem is one of the three things that keep social anxiety going, and it gets less attention than the dread before or the self-monitoring during, though it is where a lot of the damage is done.
What the post-mortem is made of
It feels like reviewing what happened, but what you are reviewing is your experience of the event, which was filtered, in real time, through anxiety. During the conversation, half of you was watching yourself from the outside, checking whether you were going red, whether that was too much, whether the pause was too long. That watching produced a picture of how you came across, and the picture was drawn by fear.
The post-mortem replays the picture, not the evening. It searches it for evidence of failure, and because the picture was drawn by fear, the evidence is always there. Then it files the result as what happened, and the file becomes the memory, and the memory becomes the dread before the next one.
Why it feels so productive
People defend the post-mortem to me. It helps them learn, they say: if I go over it, I will see what I did wrong and do better next time. That belief is what keeps it running.
All the post-mortem produces is a more detailed catalogue of a distorted picture. You cannot learn to do better at something by studying a version of it that was drawn by anxiety, any more than you could improve your driving by studying a film of a crash that never happened. What you learn from is what actually occurred, and the post-mortem is precisely the thing that stops you ever finding out.
Seeing what actually happened
This is why, in the CBT I do for social anxiety, we use video and role-play early on. You do a short, ordinary social task, and we record it. Then you watch it, and for most people this is the moment something shifts. The blushing you felt as a furnace is barely visible, and the pause you experienced as a chasm lasted a second. The person on the screen looks, to your surprise, fine.
That gap, between how you felt you came across and how you actually appeared, is the whole problem in miniature. The post-mortem lives in the gap. Seeing the real thing, even once, makes it much harder for the replay to claim it is showing you the truth.
Stopping the replay
Once you can see what the post-mortem is doing, we work on not doing it. This is a form of response prevention, and it is harder than it sounds, because the replay feels involuntary and important. The practice is to notice it starting, to name it as the post-mortem rather than as reviewing, and to decline to follow it. Not to argue with it, which just keeps it going, but to let it go unanswered and come back to the present.
Alongside that, we run experiments in real situations, dropping the safety behaviours one at a time, so that you gather evidence about what actually happens rather than what you feared.
What changes
People notice the post-mortem loosening in the first few weeks, usually once the video work has happened. The dread before shrinks after that, because there is less catalogue to dread from. And the world starts to widen, not because you have become extroverted, but because the evening no longer costs three days.
If you recognise the fifth replay and the joke that gets worse every time, there is more on how I treat social anxiety, including the video work that most people dread and most people describe afterwards as the turning point.