Not all is roses in the enterprise software world. Businesses will often refuse to buy from companies they regard as likely to disappear in a few years-- unfor

Not all is roses in the enterprise software world. Businesses will often refuse to buy from companies they regard as likely to disappear in a few years-- unfortunately, most startups tend to get lumped into this category by default. The burden of proof is on you to prove otherwise. (This is rational behavior on their part, by the way.) Managers are naturally conservative. They know that signing a slightly too-expensive deal with, say Oracle (to pick a random example), won't get them fired-- but signing a deal with some startup that fails to deliver surely will end their career. Finally, there's the problem that because sunk costs in existing systems are often high, it can be hard to break into a market. This is what keeps COBOL and even FORTRAN chugging away in corners of the world. Costs might be lower a few years after replacing these systems-- but then again, they might not. It depends on who you get to put in the replacement, and if you have trouble telling the good developers from the bad, maybe you are better off not touching it. Building and deploying is one thing. Selling is quite another, and it's not easy or cheap. Forget about word-of-mouth and friends-of-friends. Do enterprises demand a higher standard of testing? I think it depends on what your product does. I work in enterprise storage, so for us there absolutely is a higher standard of testing. I can't speak for other applications, though. Things like Yammer and Salesforce are practically the exception that proves the rule; the very rare app that needs to scale multitenant across every company in the world. Yeah, but that's where the fun is for engineers. And in time, companies like these will reduce the need for so many companies to build custom solutions.