There seem to be lots of slightly different but similar processes in these buttons. In some I have found there is a large volume of conductive material on the back of the button. The failure mode on these seems to be degradation of the surface layer (oxidation, lost of softener, who knows) and removal of the surface layer is an effective and long lasting cure. One others the problem is too much material transferred from the button to the PC board, lowering the resistance between the traces constantly. How high it needs to be varies depending on their readout strategy and the deposits causing the problem may be a long ways from the contact point.
Some of these buttons allow very little travel so adding a new pad on top of the existing ones is problematic.
Another suggestion that can sometimes work is installing a carefully placed and insulated snap disk from that style of switch. I like the idea of electrons flowing through continuous metal instead of depending on the contact between thousands of tiny particles, but the geometry on this approach is fiddly and doesn't often work.
For those putting foil or other toppers on contact cement (used to install laminate countertops) is strong and remains pliable. I can't speak to its adhesion to the "rubber" used in these buttons, but with proper cleaning it probably works. Surface prep is everything with any adhesive.