I want to do an install with an illuminated PCB plus some accent NPXL. Since I want to have Blade ID, the Illuminated PCB gets its data from Data2 (Data 3 is used by the Blade Present pin). But it obviously has no Data Out, so I would have to wire all the accents and then finish on the PCB. But the PCB gets its power from the same MOSFETs as the blade. I guess I could take +/- from the PCB, but I’m wondering: do I have to take power from the PCB all the way to the first LED in the line, or can I just wire it to the last on the output and power has to “direction”, only Data does?
You can power a strip from the end, no problem. 
 Well there is one potential problem; the draw from the strip will cause the GND to fluctuate a bit at the other end of the strip, and that fluctuation becomes noise that can potentially effect the quality of the data signal. The longer the strip is, the more likely it is that this would become a real problem.
This. Power runs parallel branches all over the place. Data is serial.
I typically run battery+ direct to PCB, then use the multiple positive pins on it as a distribution hub, so yeah, coming “backwards” through accents compared to the data direction.
#define SHARED POWER_PINS and you should be good to go.
I was going to do that, but the issue was that I’m using 5050 strips as accents and would feed power on the “last” led while feeding data from the “first”. As proff said, might add some noise, but the total strip is about 5 elements so nothing too terrible. I need to procure some 3535 strip with higher than 144leds/m density.
Augmenting by backfeeding a strip (power at both ends) is not totally un-common. With a data resistor, my money is on the data working fine.
That’s how @MeatySmurf said he wires his up, too. Having just started doing this, what are the advantages of using the pads on the PCB?