First thing: Didn't their API share those strings? Their API must enlist the values of the errors that come from servers.
Secondly, I would write it a bit differently, like this,
if (!responseString.Contains("Failed")) {
} else {
var errorMessage = responseString.Split('#')[1];
}
I would still recommend reading their API guides and documentation to learn more errors and how they are returned.
Just my own opinion, I believe using error codes is a better way to guide the clients about the errors instead of sending the strings.