Getting it right in camera is a good rule of thumb, but it's not always feasible or even possible - and seems to be going out of fashion the more powerful post-processing software gets. Oh and a Colorchecker is a huge timesaver to get the overall skin tones right!
Google "skin by numbers," there are a couple of good articles about that. Drop an eye dropper in each of the affected areas. Unlike some tutorials do, you don't have and shouldn't convert the image to HSL (lossy) unless you work in HSL all the time. You can view the eye droppers in HSL without converting color spaces (click next to the RGB figures in the Info panel).
Then slap a curves layer and start working on each of the channels. It's tedious, and a tweak in red channel shadows might result in a another tweak in blue midtones, which might cause another tweak in green highlights... You get the idea. Photoshop's coarse (low resolution) curves system isn't very helpful, and Lightzone's approach is much better, but that's another story. But it eventually gets the work done.
What Larry above suggested from Amy Dresser's tutorial sounds like it might be workable, if inelegant option to do the same as above with multiple curves layers instead of just one. I'll try that next time I have that issue. Pixel peepers would balk at the errors produced by the multiple adjustment layers
Finally, believe your eyes. Take a break and come back later to see if it still looks the way it's supposed to. You probably don't want to go for a purely neutral look. As Slobodan suggested, most people want a natural look, and shadows have different color temp than highlights, especially in mixed lighting.
Then there are some who want a clinical look, just like there are architectural photographers who insist on "fixing" converging verticals although that's how we see them in real life...