OP here - me, too :)
Question: are you including every tweet in your analysis, or do you ignore some as being irrelevant? I mentioned the example of people complaining about spam from public whois records; if you mark such tweets as negative, I could see that explaining a large part of it.
(I ignore them as I don't see it as being directly related to a host's performance, even though I know there's an argument to be made about public and "private" whois records).
FWIW I also ignore positive tweets that consist of short generic "great, thanks!" or "Yay, awesome!" - I've found that they most often tend to be talking about some sponsorship/promotion or meetup/party the host has held.
Do you have a link to your site? Be curious to look at what you're doing.
My data doesn't agree with what he's saying. GoDaddy's reputation is on par with the majority of the EIG brands. A Small Orange was a good company post acquisition by EIG for a while. I wrote about that specifically here: http://reviewsignal.com/blog/2016/01/19/the-rise-and-fall-of... basically it went to hell once EIG management got involved and the original people team/staff left and it got merged into the conglomerate.
I'd be curious to see his data as well and why it his conclusions seem so different.