> If it makes them looks .1% better and improve signups by the same amount they're golden
I have trouble imagining that any sort of measurement tying signups to the logo would have a margin of error smaller than that. I assume this is just a made-up number to signify how much they care about slight changes, but I think the general point that they realistically would have trouble getting any amount of useful signal from slight tweaks like that still stands.
> I assume this is just a made-up number to signify how much they care about slight changes
Yes it is, I have no idea what they monitored but I'm sure they didn't chose randomly.
> they A/B tested their logos for months
I doubt it; I could instantly spot jagged edges of their design, it looks more like a rush job.
A/B tests imply that people have different opinions.
They probably spend millions in design per year and have to keep their designers team working. At that scale nothing is left to chance, they A/B tested their logos for months &. If it makes them looks .1% better and improve signups by the same amount they're golden.
> I only noticed those details when they were explicitly pointed out.
And that's exactly how it should be, good design is imperceptible, bad design sticks out like a sore thumb.