Compression reduces bytes by changing how an image is encoded. It does not guarantee a particular percentage saving, and an already compact source may not become smaller under every setting.
Image Compressor offers JPEG and supported WebP output with a chosen quality. Review the generated file rather than assuming that a quality number is a file size percentage.
Quality and dimensions are separate controls
A 1200 by 800 picture remains 1200 by 800 after this operation. Lowering quality can reduce bytes while leaving that same grid of pixels in place.
A noisy photograph, a flat illustration and a screenshot can produce different sizes at the same quality. The content affects compression, not just the entered number.
JPEG flattens alpha onto the selected background. WebP can retain transparency when the host encoder supports it. Choose the codec based on the receiving system and the asset's needs.
Use a sensible comparison area
Inspect small text, hair, edges and smooth gradients at normal display size. Those areas often reveal whether the chosen compression has removed too much useful detail.
If the result is still above a strict limit, resizing may be more effective than repeatedly lowering quality. A large image contains more pixels to encode, even when each pixel is represented compactly.
Batch compression applies the same quality choice to every file. Their sizes will differ because their content differs. Download bytes are reported separately so an oversized member is not hidden by the overall ZIP.
For an exact byte ceiling, try the Target File Size tool and read its limits. For a general website optimisation pass, keep the best source and choose the smallest output that still looks suitable at the size visitors will actually see.