A PNG that is far larger than you expected is usually not a broken file. It is a lossless format doing exactly what it was designed to do, applied to an image it was never meant for.
Lossless means nothing is thrown away
Decode a PNG and you get back precisely the pixels that went in. It saves space by describing those pixels more compactly, not by discarding any. The practical consequence is that there is no quality setting: any option that claims to compress PNG by reducing quality is either ignored or quietly converting the file to something else.
So when a PNG comes out the same size or slightly larger after "compression", that is the honest outcome. The encoder had nothing to exploit that the original encoder had not already exploited.
The usual cause: a photograph stored as PNG
PNG is built for flat colour and sharp edges. A photograph is the opposite: millions of subtly different values, no repetition to find, and nothing for a lossless encoder to compress. The file ends up several times larger than the same photograph saved as JPEG, with no visible benefit at all.
If the image is a photograph and does not need transparency, converting it to JPEG - or to WebP - is the single biggest saving available. WebP or JPG compares the two.
When the cause is dimensions
A screenshot taken on a high-density display can easily be 2500 pixels wide. If it is displayed at half that, it is carrying four times the pixels it needs. Halving both sides quarters the file, and unlike a quality setting this damages nothing.
What actually reduces a PNG
- Reduce the dimensions to the largest size it will be shown at.
- Convert to WebP, which keeps transparency and compresses far better.
- Convert to JPEG, if it is a photograph and transparency is not needed.
- Re-encode with an optimiser that tries many filter combinations. Saves a little on some files, slowly.
When keeping the PNG is the right answer
Screenshots, diagrams, charts, logos and anything with crisp text. JPEG handles those badly - visible fringing around letters, mush in flat areas - and for that kind of image PNG is often the smaller file as well as the correct one.
A tool that tells you the file is already close to optimal is more useful than one that invents a saving. Compress a PNG and check the before and after sizes yourself; nothing is uploaded, so trying is free.