Disclaimer: I've worked as an EA for about 10 years, on top of another 10 years of solutions architecture and applications development. I've worked for a bunch of private and public organisations.
It's so easy to do EA badly. If you treat it like a lead or senior architect setting the technology strategy, then you're missing most of the benefits (and this should be more the CTO's domain anyway). If you do ivory tower strategy that nobody ever reads, because it's full of stuff nobody can relate to, then that's pointless and you're unable to communicate, which is the primary purpose of an architect. If you have a small outfit where those setting direction communicate well with those designing the business and technology, then you don't really need an EA.
An EA is an architect for the _business_, not (only) for the technology. The reason a lot of people get into it from technology is it shares a lot of the rigour and approach of architecture, however it is important to note it is NOT primarily a technical role, nor should it be.
A successful EA will understand the business strategy, i.e. where the company needs to be in X years, and understand the existing landscape of roles, skills, processes, technology and so forth. Their primary purpose is to define the change the business needs to go through, in each of those areas, in order to fulfil the business strategy.
A C-level execs set the strategy. The EA transforms that strategy into changes that need to occur within each level of the business, to make their vision possible. The individual specialist areas (from HR to tech to whatever) work with the EAs to determine how to make those changes happen in that timeframe.
Done well it's an incredibly powerful tool and is the mechanism that connects the "controls" to the "engine". Done poorly it can fail for any of the reasons above, from people who see it as having a purely technical remit to those who sit around in Archimate all day making models nobody will ever use.