A ten minute set sounds like a person talking, and it is almost always a designed structure. The parts are consistent enough that comedians discuss them by name.
The premise is a claim, not a joke
Every bit opens by establishing a proposition the audience can accept: something is annoying, something is strange, something is not what people pretend.
The premise carries no laugh itself. Its job is to get the room to the same starting position so the punchline has something to overturn.
A premise that needs lengthy explanation is usually the wrong premise, which is why comedians strip setups down over months of performance.
The act-out does the heavy work
Rather than describing a person, a comedian becomes them briefly, adopting a voice and a posture for a few seconds.
Showing is faster than telling and much funnier, because the audience recognises the imitation before any joke has been made.
Act-outs also break the rhythm of a person standing and talking, which matters over ten minutes more than it does over two.
Tags extend a laugh already earned
After a punchline lands, a tag adds another twist to the same setup without needing new context.
This is efficient, since the expensive part of a joke is the setup and a tag reuses one that has already been paid for.
Strings of tags produce the accelerating sections that make a set feel like it is gaining speed towards the end.
Callbacks make ten minutes feel whole
A callback returns to an earlier joke at a later point, usually in a context that has nothing to do with the original.
The laugh comes from recognition rather than surprise, and it rewards the audience for having paid attention.
It also converts a sequence of separate bits into something that appears designed, which is the main reason closing callbacks are so common.
The order is the last thing decided
Individual bits are built independently, and arranging them is a separate problem solved after each one works.
Openers need to establish who the comedian is quickly, so the strongest material is rarely first.
Closers have to be reliable rather than adventurous, because a set is remembered by how it ended, which is why comedians keep a closer long after retiring everything around it.
New material is then slotted into the middle, where a failure is buffered on both sides by bits that are known to work.