Very frequently, the right answer for an MVP is something like "an e-mail newsletter" or "a Salesforce extension" or "a CSV file" or "an IDE plug-in" or "throwing a pizza party for your target market and performing the service for them yourself". I've seen startups charge $3K+/month to dump a CSV file on a client's FTP server. Go where the user can most conveniently make use of your product; for a lot of businesses, that is neither app nor website.
I agree. In many cases, don't start with either (app or site). In other cases (or in cases where you've grown up to the point of needing an app or site), start with one that is right for your user.
