5% different, almost everyone is online.
But, thinking about your oft-quoted blog post about advertising bring a cancer, I guess if the top ad spenders cut themselves out entirely, then the bidding system could result in the runner-up bidder finding their ads are now almost arbitrarily cheaper.
We at best get 70% participation in voter representation. Online wise, I'd wager we get at best 20% "public" participation in terms of who bothers to participate online as opposed to lurking. And that 20% is spread amongst hundreds of subjects.
And we know these aren't created equal. The internet is disproportionately on the younger side, is very slightly biased by "middle class", educated workers, and the gender demographics vary site to site, despite being overall even (e.g. disproportionately male on Reddit, female on Pintrest). There will definitely be a different resuts online compared to making physical surveys.
I can easily imagine this to alone be responsible for wiping 84% revenue.
Real world has a different political distribution than the Internet. "Politically toxic" on-line in particular is a knee-jerk reaction that is great at generating consistent revenue streams for publishers and social media on-line, but doesn't translate well to how the entire population of a country actually thinks or votes in the real world.