OK, so let's not get hung up on technicalities as I am often inclined to do. Yes, a coating would be on top, and if it is scratched it MAY leave a visible mark, however it may not. What type of coating it is - that is, whether it has optical refraction qualities (i.e. causes the light to bend in any way, changes the wavelengths of light that pass through versus get reflected back as was referred to by another member earlier), will determine whether if "scratched" or otherwise removed it will leave a visible scar. I don't think (though I may be wrong), that the oleophobic coating would leave a visible scar if scratched off of a portion of the glass surface. Oleophobic coatings are in the range of 2 - 10 NM in thickness, which in itself may be so thin that it's virtually undetectable to the naked eye. Also, the coating may be technically "on" the surface, however remember we're talking about a surface that is full of pits and islands. So what seems like a flat surface to you is actually very rough. The "coating" will actually "fill in" the tiny surface imperfections and will therefore be thicker (we're talking NM here), than where the coating is able to adhere to the islands (if at all). In order for the oleophobic coating to work properly, it has to smooth out the imperfections as much as possible so as to remove any that the oil can latch onto, hence the oil then beads up and looks like it's not there - mostly. Here's a few pics of a similar process using a hydrophobic coating (reject water). You'll see the (simulated) surface of the glass, then see the coating applied over a portion of it, then you'll see actual working versions of uncoated and coated glass. The idea is that the water doesn't want to adhere (attach) to the glass. I am simply trying to get across the point that for our purposes, we think of most (if not all), scratches on the screens of our phones to be scratches to the actual glass surface. I would bet that 99.98% - 100% of all scratches fall under that realm.