We discuss here optical losses in silicon heterojunction solar cells and strategies to minimize them. Optical losses originate from most non-crystalline-silicon layers involved in the solar cell. A breakdown for typical values is shown evidencing that suppressing absorption from the front amorphous silicon layers gives the largest gain. Other losses are interdependent, and reducing absorption from one layer in the infrared part of the spectrum boosts absorption from the other layer. The use of nanocrystalline silicon layers in lieu of amorphous silicon enables a reduction of the absorption at a given thickness but thicker layers are seen to be necessary, reducing the eventual optical gain in optimized devices. For the front transparent conductive oxide (TCO), infrared light saved from absorption in the front TCO will be shared between absorption in the c-Si wafer and in the rear electrode, and a share will also be outcoupled from the device through reflection (and transmission in the case of bifacial devices). The cell architecture will therefore dictate how much of the current saved from parasitic absorption in the front TCO will eventually benefit to the device. Using a thin ITO combined with silicon oxide is a route to provide similar electrical performances with reduced indium use and a slight cell-efficiency boost. Switching to a high-mobility Zr-doped indium oxide layer enables to use as thin as 35-nm-thick layers with still low series resistance, which outperforms optically in a cell configuration but yields similar results in a module configuration.