Manual indexing can be done by entering cell lengths and angles and you can then manually change lengths using the up and down arrows next to the length or angle to increment or decrement the value. This can be a hopeless task with a complex pattern and a low symmetry cell, but can work well for indexing cubic, hexagonal or tetragonal cells, where if two peaks can be indexed as a h00 and a 00l reflection, the lattice has then been defined. On occasion more complex cells can be indexed if there are low index peaks along each axis. I have also had success with this type of indexing where electron microscopy can identify a 2D unit cell and only a third lattice dimension and possibly an angle needs to be determined.
Note that autoindexing requires a minimum of 20 peaks, so for minor phases, this is not usually possible. Database searching, for example with the ICDD or COD databases may help identify the phase if already known. For new phases, manual indexing may be the only hope.