Ah, thank you for calling me out on that; I completely misremembered the anecdote.
It wasn't total spend; it was online campaign spend. "Chaos Monkeys" cites a Bloomberg report on an internal Facebook memo that indicates the Trump campaign ran six million different ads on FB during the campaign and the Clinton campaign ran 1/100th of that amount. So targeted ads were involved, but the targeting approach was very traditional: pay a bunch of advertisers a lot of money to hand-tune ads, see how they perform, re-tune, rinse, repeat. The spend on Cambridge Analytica as a ratio and the effect it had on the total process were both minimal; CA didn't prove to be the "voter whisperer" that the owners made them out to be, and in the long run, the fact that they exfiltrated a bunch of private content from Facebook's datastores isn't as interesting as how the Trump campaign took advantage of the data in Facebook's datastores using the tools Facebook legitimately provides.
One feature the campaign did (according to the author of Chaos Monkeys) find useful was "Lookalike Audiences," which is nothing fancier than crawling the social media graph and expanding an initial targeted ad along friend networks (i.e. if an ad seems to be resonating with you, Facebook's own algorithm, if the advertiser has enabled the feature, will try pushing the ad to your friends and so on). In that sense, the data Facebook collected facilitated a Trump victory, though it wasn't anything more dangerous than the social graph itself... And I don't think the EU is proposing we ban social media or collecting networks of friends at this time.
... though maybe they should? You can do a lot of damage with the information people voluntarily share about who they associate with, if you collect enough of it.
> I don't know how you can break apart Americans disliking women in general and Americans disliking Hillary particularly
A good and fair question. So it turns out one of the largest blocs of votes in the 2016 election was various flavors of Christianity, and they generally chose to vote for a known womanizer and divorcée (with Protestants and Catholics, in particular, voting for Trump by a wide margin over Clinton). This would be considered curious behavior, except scraping the surface only a tiny amount reveals that they are almost 100% unified against the concept of women in a leadership position; some have structural taboos against it, and to some it is an existential threat in the category "God will strike us down for our hubris" because it goes against their notion of a cosmic order. It's ugly and I wish it were not so, but I think most political pundits wildly under-estimated that effect because, as the first woman to be nominated by one of the two major parties in America, their prediction models had no data on what effect it would have. I agree that the fact she already had a political service history that could be criticized (vs. her opponent with no such service) was also a factor, but I don't think it was as large a factor as the voters who turned out with fear of actual divine retribution in their hearts due to their religious beliefs.