Compression looks for a shorter way to describe the same data. Run-length encoding replaces a repeated run of the same symbol with just a count and the symbol itself, which helps when runs are long but can backfire when data changes constantly.