Examining this video was an interesting exercise. I was able to get this to work as he shows, but had to watch a few times. What is curious to me is that I was able to get the needed face snapping behavior without any of the checkboxes activated that he recommends in the video, which confuses me. Just to be clear, there is nothing special about the whole Object Mode/Edit Mode situation you describe. It is true, the eye is in Object Mode while the head is in Edit Mode, but the relevant point is that the video demonstrates *Face Snapping* while using a checkbox called **Project Individual Elements**. According to the tooltip, this feature allows the snapping elements to interact with other objects. So that explains how he's able to bridge the gap between Edit Mode and Object Mode. Except that it doesn't because as I said, I was able to get it to work without the Project Individual Elements option checked. So this is a mystery to me. Maybe someone else can shed some light.

I thought I would point out that I use a different technique to get eyes and eyelids to meet, and I think it's superior (if only because it requires no vertex tweaking) than the one in the video. I create a vertex group, and it contains only the vertices from the edge ring which defines the inner rim of the eye lid. Then I create a Shrinkwrap modifier for the head, and point it to the eye object as its target. There is a **Vertex Group** field for the Shrinkwrap modifier. Point it to the vertex group you created for the eye lid rim, and it will instantly snap to the eye, while leaving the rest of the head mesh alone. This achieves the same result without the need to manhandle every single eye lid vertex in order to get it to meet with the eye ball geometry. It also doesn't miss.