You can chair a review with twelve people in it, say the difficult thing, disagree with someone senior, and be completely fine. Then the meeting ends, people drift toward the coffee machine, and something drops out from under you. Same people. Same room. Two different versions of you, forty seconds apart. And later that day the same person who handled the review can lose their voice with the checkout clerk.
If you have ever typed some version of "confident at work but socially anxious" into a search bar at eleven at night, you already know the confusing part. It doesn't look like a lack of nerve. You have plenty of nerve, in one specific place. So what exactly is switching off?
Confidence is local. It doesn't travel.
The first thing to know is that confidence was never one general substance you either have or don't. In psychology the closest precise concept is self-efficacy, Albert Bandura's term for your belief that you can handle a particular kind of situation. Its defining feature is that it is domain-specific. It builds in the exact conditions you repeat, and it stays there.
So fifteen years of doing your job well buys you a large, real, well-earned confidence in doing your job. It buys you almost nothing at a birthday party where you know two people. That isn't a flaw in you. That is how the mechanism works for everyone. The unusual part is only how far apart the two domains have drifted.
A role does half the talking for you
Work situations are what researchers call structured. The rules are visible. You know what the conversation is for, what you are expected to contribute, roughly how long it lasts, and how it will be judged. There is a purpose sitting on the table between you and the other person, and both of you are looking at it rather than at each other.
Unstructured social situations remove all of that. No agenda, no task, no ending, no criteria. Nothing is on the table, so the only thing left to look at is you. Ambiguity is fuel for anxiety generally, and this is ambiguity of the most personal kind: you cannot tell how you are doing, because there is no such thing as doing well here in any measurable sense.
A role does something else too. It gives you a reason to be present that is not simply yourself. You are not standing there as you, you are standing there as the person who handles that thing, and that is a legitimate ticket into the room. Take the role away and the ticket goes with it.
The job wasn't making you confident. It was saving you from having to justify being there.
What is actually being evaluated changes
At the centre of social anxiety, in every mainstream model of it, sits the fear of negative evaluation: the expectation that other people are appraising you and that the verdict will be bad. Clinical models such as Clark and Wells' describe what happens next. Attention swings inward. You start building a mental picture of how you must look from the outside, and then you monitor that picture instead of the actual conversation. Which is why you can leave a five-minute chat with no memory of what the other person said, but a vivid image of your own face.
Work protects you from that loop in a specific way: it gives you evidence. The report either worked or it didn't. The client either signed or didn't. There is a scoreboard, it is external, and you can check it. Social life has no scoreboard, only faces, and faces are ambiguous enough that a worried mind can read anything into them. So the monitoring never resolves, and it follows you home. Replaying the conversation afterwards, rewriting one sentence you said, is the same process running late. Researchers call it post-event processing, and it is one of the most reliable markers of the pattern.
Why competence doesn't carry over
Here is the part that tends to land hardest. At work you are valued for what you produce. That is a fair deal and you are good at it. Socially, nobody is buying anything. The question underneath is quieter and much worse: whether you are worth having around when you are not being useful.
If your sense of your own worth is built on contribution, then a setting with nothing to contribute takes the floor out. And it explains the direction of the split. It isn't that you can't talk to people. You can talk to people all day, expertly, as long as there is a reason. Remove the reason and the whole thing goes quiet.
This is the difference between confidence and self-worth showing up in one life, and it is worth being clear about which one is short. Confidence and self-esteem are two different systems, and stacking more achievement will keep topping up the one you already have.
What this changes
Two things follow, and they pull in different directions, which is why this gets misdiagnosed so often.
The first: since confidence is domain-specific, it can only be built in the domain you actually want it in. No amount of professional success will quietly convert into ease at a dinner table. The reps have to happen where the gap is.
The second, and the reason "just get out there more" so often disappoints: reps don't fix a worth problem. If what is running underneath is a question about whether you are allowed to take up space at all, then more parties mostly give that question more places to ask itself. The evidence never lands, because the fear was never really about the parties.
Which is the honest answer to why you can be so capable and still feel like this. Nothing is broken in your social ability. Something is unresolved in what you believe you are worth without a job title attached, and that is the layer worth working on first. Once it shifts, going out there stops being a test you are trying to pass.
Common questions
Is it normal to be confident at work but anxious at parties?
It is common. Confidence builds in the exact situations you repeat and does not automatically transfer to situations with different rules. Work hands you a role, a purpose and clear criteria. A party hands you none of those.
Why does my confidence disappear the moment work is over?
Because what is being evaluated changes. At work you are judged on output and there is evidence. In unstructured settings there is no task and no scoreboard, so the thing under evaluation becomes you, and attention turns inward.
Can you be successful and still have social anxiety?
Yes. The two live together in a lot of people, and sometimes the success is partly powered by the anxiety. Being good at something is not the same as feeling allowed to exist without it.