Not fucked up, it should just be limited to comparing similar-spectrum light sources. It's a quick back-of-the-napkin way to get the lay of the land, like Vero vs Cree 3500K 80 CRI, or whatever. Gets you in the neighborhood.
What we care about foremost is photons per second, since that's how plants photosynthesize. One photon at a time. The onlu way we have to figure that out is to reverse the lumens weighting using the spectrum charts from manufacturers, to get actual photons popping out. If you look at Supra's charts you'll see something like this:
View attachment 3731476
See that umol/s/W? That's the important bit, more so than lm/watt. It's the number of photons per second per watt, but actually figuring that out accurately is pretty hard. Ideally you take a light, send it to a lab, and have them tell you. In this case it might have been actual Cree-provided numbers, I don't remember, it's buried somewhere here.