What Base32 actually does
Base32 reads the input five bytes (40 bits) at a time and splits them into eight 5-bit groups, mapping each group to one of 32 printable characters. When the final block is short, = padding fills it out to a multiple of eight characters.
The cost is size: output is about 60% larger than the input, versus 33% for Base64. The benefit is robustness — the alphabet omits 0/O and 1/l confusion and ignores case, so it survives transcription, phone dictation, and case-mangling middleware.