What intrigues me most about your article is that it begs another question. How much testing is enough? Based on your account of the situation I don't think you

What intrigues me most about your article is that it begs another question. How much testing is enough? Based on your account of the situation I don't think you tested enough variables. Were the premium features just something you threw up as an attempt to monetize the site or did you really think through the needs of someone who is willing to pay you? Stepping through your site there are lots of suggestions I would make (which are beyond the scope of this comment), but at what point can one say he's done enough testing to say "I'm done with freemium". That's the question I'm interested in.