The real tell-tale in those buttons are their styles rather than the colors. If you took the solids and lined them up as palettes, it would be a lot more difficult to parse. The infographic showing the emotion colors correlating to their brands does more to prove that big and bold is what matters, not the hue (and honestly it looks like someone just pasted all the logos they could think of, regardless of if they supported the meaning/emotion). I'm not even going to touch the gender-based one. I tried to reach the single URL that Kissmetrics lists as their proof but it no longer works (also why don't they have HTML versions of their data?).
At the end of the day, it really comes down to how color is used. There are lots of great books and blogs on color psychology, and I'd suggest looking into those instead.
Fast food seems to really prefer warm colors (red, orange and yellow) - think, McDonalds, Wendy's, Subway, KFC, Dominos, Burger King, Jack in the Box, Sonic etc. This correlates with the post, where red = energy + creating urgency while yellow = optimistic and youthful.
Fast food: making customers energized and in a rush since the color red.
Alcohol for some reason really loves pure black or pure white (I'm thinking to tv commercials with black backgrounds and the slow-motion alcohol pours). Again, great correlation here in the post where black = luxury/power.
But thinking to my own personal experience with design, Leo's quote stands out to me:
"Despite all the studies, generalizations are extremely hard to make. Whatever change you make, treat it first as a hypothesis, and see an the actual experiment what works for you."
It seems to me like a blue/orange combo has become one of the most-used color schemes for startups (correlation to movie posters? http://www.slashfilm.com/orangeblue-contrast-in-movie-poster...)
Really? Crap... I haven't really noticed, and it probably doesn't matter, but we're blue/orange heavy. For what it's worth, we went with this design 3 or more years ago, so at least we probably can't be considered copycats. :-)
I think it is a bit of a reach to immediately jump to concluding red outperforms green. Cohorts in a/b testing are not always equal (even if the traffic split is 50/50).
Imagine situations where 30% of a has 70% more mobile users than those of b. The green used for b is significantly less noticeable on mobile than desktop.
Can we still reasonably conclude that color defined the conversion difference?
Wouldn't this be statistically impossible (i.e. extremely unlikely) given a big-enough data-set in a properly designed A/B test platform?
Of course, given IBM's history and nickname of "big blue" I guess a lot of people would default to associating "blue" with "enterprise software".