Making Files Smaller: Compression, Conversion and What Actually Works

Lossless compression looks for repeated patterns and stores them once. It is why a TXT or CSV shrinks heavily, why a DOCX is already a ZIP archive before you touch it, and why PNG beats BMP. The output can be restored byte for byte.

It also has a hard ceiling. A JPEG is already the result of lossy compression — the redundancy was removed when the photo was saved — so archiving it as a ZIP gains almost nothing, often under one percent. The same applies to MP3, MP4 and any other lossy file: you cannot re-find patterns that are no there.

When a file must get substantially smaller and it is media, the answer is re-encoding at a lower quality setting. Images: lower the JPG quality, or move to WebP/AVIF, which reach the same visual quality with fewer bytes. Audio: drop from 320 kbps to 128–192 kbps, which most listeners cannot distinguish on phone speakers. Video: re-encode at a lower resolution or bitrate, which dominates every other saving because video is the biggest thing on any device.

The rule that keeps this safe: re-encode once, from the original. Each pass throws away a fresh layer of detail and compresses the artefacts of the previous one. Chain three conversions and the result is visibly worse than a single careful one — for a larger file, because artefacts are harder to compress.

When you need to email a folder or reduce the transfer size of many small files, pack them into an archive. ZIP is the universal choice — it is built into every operating system. 7z and RAR typically achieve 5–15% better compression on documents and logs; TAR.GZ is the convention on Unix systems. None of these will meaningfully shrink a file that is already compressed, such as a JPEG, MP3, MP4 or another ZIP.

For documents specifically, the format itself often offers the biggest win. Exporting a Word document with embedded photographs to PDF usually shrinks it substantially, because PDF's image handling and font embedding are more efficient. A presentation full of full-resolution photos is the classic case: the same deck as a PDF is routinely half the size.

Do not convert a lossy file to another lossy format hoping for savings — the size change is usually negligible and you have spent quality for nothing. Do not convert to a lossless format if size is the goal; that always makes lossy sources larger. And do not expect any tool to shrink an already-compressed file by a meaningful percentage: the information density is fixed by the format.

If a file genuinely must be small, the levers in order of impact are: remove what is not needed (unused pages, extra tracks, full-resolution originals), reduce resolution or duration, then re-encode once at an appropriate quality. Everything else is rounding.