However, my hunch is that many startups fall into the former category (jack of all trades) almost accidentally because they don’t have the will/vision/stubbornness/whatever to buckle down and do the latter. That is, they are not making an explicit choice, which may ultimately not be in their best interest.
The reason is that delivering features people ask for is the path of least resistance. Not delivering them requires you to essentially ignore (or at least gracefully put off) huge obvious feature requests and focus diligently on stuff that seems much smaller, and to the untrained eye, perhaps trivial.
And that’s the key. Are these small things really trivial or are they part of a larger product vision where you end up with a truly polished product? It’s often hard to tell, and sometimes really a probabilistic bet. You really never know if you can nail a product experience until you do.