That depends. I worked at a company where we revamped the entire checkout process including putting labels inside the input field and we had a significantly higher conversion rate than using the standard checkout forms one would find on a typical checkout process we had before.

Part of that was simplifying the look of the checkout, which includes putting the labels inside of the inputs.

Not to mention some very notable websites use this design choice, including Square, Twitter, and Apple.

Comparing conversion rates of a complete before and after revamp won't tell you anything about the effectiveness of individual components of the revamp. You'd need several A/B tests or a gradual series of changes to tell you whether labels inside or out actually made a difference overall.

I know it's a silly argument but: do you think Facebook have made some A/B testing for their registration form ?

Because they seems to think that the placeholder as label works well [0].

[0] https://www.facebook.com/

Interesting - do you have any links to the studies you mentioned (not doubting, just curious)?

I remember 1 or 2 were from UXMovement.com and others from all over the place, usually developers who run A/B tests and report their results on the site's blog.