This methode needs one and the same object two times. The first one is the object with your transparent-holographic-fancy-whatever material and custom depth on, the second one is just a mask for the first one, it also needs custom depth activated, but “Render in Main Pass” needs to be deactivated, so you can’t see it. It just is a mask in the custom depth buffer. So if you want to place an object in your scene and activate custom depth and want it to work, you need a second object of the same kind, with the same transforms and, if their is any animation, this syncronized, and “Render in Main Pass” deactivated and “Render Custom Depth Pass” activated.
tl;dr: It’s easier in blueprints 