The company that makes them: https://www.dyaqua.it/invisiblesolar/_en/
When they talk about "sun rays", do they mean the UV portion of sunlight? Because if the material were transparent to visible light, it would be, well, transparent, wouldn't it?
I don't think so, because they say they use a standard monocrystalline siicon PV cell -- which are (generally) crappy for UV scavenging.
I would guess the material atop the solar cell is either slightly porous or made of a material that is mostly transparent to the right spectra of light, resulting in acceptable losses.
Our eyes see, what, 200-600 nm? Removing the 200-600 nm bit leaves you with approx 50% of the energy left between 600 and 1100 nm, glancing at [1]
This is all VERY back of the envelope... but I think these lose around half of the energy you'd get out of conventional panels.
[1] https://www.researchgate.net/publication/303914764/figure/do...
But even a ~20% reflection coefficient can make something appear "opaque" if viewed in sunlight and complete darkness behind it.
It's unfortunate they don't just have actual numbers on their website, which is a pretty terrible indicator, especially with the orientation of the panels being around 45 degrees.
[1] http://environmath.org/wp-content/uploads/2020/12/spectral_l... from http://environmath.org/2020/12/17/why-solar-panels-cant-get-...
If you have a flat roof then all the space is available, and depending on your latitude flat can be a good orientation, especially in summer.
Pompei roofs do not appear to be flat, but they do seem to be very low pitch, and the panels are curved, so the whole roof is likely in play.
Back to your main point, effeciency in itself doesn't matter. If you have enough space, you just install more. That drives up the installation cost though, and things like cleaning them etc.
Of course ultimately everything is space constrained, but generating 11% is better than 0%.
For the benefit of the audience: if you imagine picking a point on that curve and drawing a square under and to the left of it, a solar panel captures only that energy from the orange region. Photons with longer wavelengths aren't captured: not enough energy to push an electron up the bandgap. Photons with shorter wavelengths have more and more energy to push an electron up, but can only capture 1.9226e-19 joules per photon of that energy at 1.2eV.