"Convert" is one of those words that hides a lot of different operations behind a single friendly button. Converting a JPG to a PNG, converting a HEIC to a JPG, converting an MP4 to an MP3, converting a PDF to a Word document — from the outside, all four look like the same thing: you pick a file, you pick a target format, you get a new file. Under the hood, all four are doing completely different work, and the quality of the result depends on which category you're in.
Understanding those categories takes about five minutes and saves you a lot of "why did my file get bigger / worse / mangled" surprises.
Category one: the perfect swap
Some conversions are lossless — the new file contains all of the information the old file had, just written in a different notation. Converting a PNG to a BMP, a WAV to a FLAC, or a CSV to a TSV falls in this bucket. If you convert back afterwards, you get exactly what you started with. These conversions are boring in the best way: nothing surprising can go wrong, and file size changes are the only thing to notice.
The tell for a lossless format is that repeat conversions don't degrade it. Convert a PNG to a BMP and back a hundred times and pixel one hundred still matches the original. This is the mental model most people carry into every conversion — and it's true for maybe a quarter of them.
Category two: the shrink-and-approximate
Most everyday conversions live here. Converting a PNG to a JPG throws away information the JPG format can't represent — small colour differences in smooth areas, transparency, some fine sharpness — and produces a smaller file that looks nearly identical. Converting a WAV to an MP3 throws away frequencies humans can't hear well and produces a smaller file that sounds nearly identical. Converting an MP4 to an MP3 throws away the entire video track and keeps just the audio.
These conversions are one-way. If you convert a PNG to a JPG and then back to a PNG, you don't get the original PNG — you get a PNG-of-the-JPG, which is missing the information the JPG threw away. Each round trip loses a little more. That's why professional workflows keep an original in a lossless format and only export lossy copies for delivery.
The tell for a lossy conversion is that quality settings appear in the dialog — "quality: 85", "bitrate: 192 kbps", "compression: high". Those numbers are you telling the encoder how aggressively to approximate. Higher numbers mean bigger files and less loss.
Category three: the format-of-formats
Then there are conversions between formats that aren't really comparable — the classic being "PDF to Word". A PDF is a page-layout format designed to look the same everywhere. A Word document is an editable text format that flows to fit the reader's page size. These describe fundamentally different things, and the "conversion" is really a heuristic reconstruction: the tool reads the PDF's positioned text and lines and tries to guess what the Word paragraphs, tables, and headings would have been if the document had started as Word in the first place.
This works well for simple text-heavy PDFs and poorly for complex layouts, scanned documents, or anything with heavy formatting. The output is a best-effort translation, not a lossless conversion, and a human usually needs to spend a few minutes fixing tables and headings afterwards.
The same is true for HTML-to-PDF (works well), PowerPoint-to-PDF (works well), PDF-to-Excel (works badly on anything except plain tables), and eBook conversions between EPUB and MOBI (works well within limits). If you're in this category, judge the tool by the quality of the guess, not by the fact that it produced a file.
Category four: the container swap
Some conversions rename the wrapper without touching the contents. Converting an MKV video to an MP4 usually falls here: the actual video and audio streams are copied byte-for-byte, and only the container format around them changes. That's why it's instant, and why the file size barely changes. Converting a DOCX to a ZIP is the ultimate container swap — a DOCX is a ZIP file with a different extension, and the "conversion" is just renaming it.
These conversions are lossless and fast, but they only work if the contents of the container are compatible with the new wrapper. If the MKV had a codec that MP4 doesn't support, the tool has to re-encode it, and you drop back into category two.
How to know which category you're in before you click
Ask two questions: Does the target format support everything the source format has? Is the tool re-encoding, or just rewrapping? If the answer to the first is yes, you're in category one or four (safe). If the answer is no, you're in category two or three (approximate).
As a rule of thumb: file-size changes are your best signal. A conversion that goes from 40 MB to 40 MB is probably a container swap (safe). A conversion that goes from 40 MB to 4 MB is definitely lossy (approximate). A conversion that goes from 4 MB to 40 MB is either lossy → lossless (fine, but pointless — you don't recover what was lost) or a format mismatch (like PNG-ing a photo).
Why doing it in the browser matters here
One more thing worth knowing: when a conversion tool runs in your browser, all the categories above are exactly the same as they would be on a server. The tool uses the same libraries (FFmpeg, ImageMagick, PDF-lib) compiled to WebAssembly. The output is identical. The only difference is that the file never leaves your device — which matters most for the exact conversions people do most often on sensitive files: PDF-to-Word for contracts, HEIC-to-JPG for passport scans, and MP4-to-MP3 for meeting recordings.
So the real advice is: pick the right tool for the category (lossless where you can, quality-tuned where you can't, browser-based when the file matters), and stop treating "convert" as one operation. It's four.
