> I've more often heard of the opposite problem.

Yeah, my guess is this is the more common case for a randomly selected startup. But I suspect Casey doesn't advise a randomly selected sample of startups.