JPEG to WebP: compare size without losing what matters
When preparing pictures for a catalog or web page, the goal is not the smallest file at any visual cost. It is a lighter file that retains what the reader needs. Decide which details matter before looking at the byte count.
Keep dimensions constant for a fair comparison
Compare JPEG and WebP at the same pixel dimensions and intended display size. If one version is half as wide, you cannot tell how much of the saving came from its format and how much came from fewer pixels.
Choose meaningful areas: a label, the product outline, or a sky gradient. Inspect them at normal viewing size first, then enlarge them to examine compression effects. A good-looking center does not compensate for lost information at an edge.
Quality is not a percentage of the original
The quality setting for lossy WebP controls a size–quality trade-off. A value of 85 does not mean that 85% of the original detail survives. Lossless encoding is a separate mode, not just a high quality number.
This conversion path produces lossy WebP with an adjustable quality setting. Inspect the actual downloaded image and then compare its size. The format name alone is not a promise that the result will be smaller.
References: Google WebP — Quality and lossless encoding are separate controls
Measure this picture, not a general promise
A WebP output can be larger than its source in some circumstances. If conversion offers little saving or damages an important detail, keep the original or try an appropriate setting rather than forcing the same decision on every picture.
Before replacing assets in a working library, open a sample in its destination application. Retain originals outside the export chain; a compressed publishing copy should not become the source of every future edit.
References: Google WebP — File size, format features and limitations
Worked example: is the saving worth a replacement?
For illustration, suppose a JPEG is 240000 bytes and a visually acceptable WebP is 132000 bytes, with unchanged dimensions. The difference is 108000 bytes, so the saving is 108000 ÷ 240000 × 100 = 45%. These invented values demonstrate arithmetic, not a converter benchmark or promise.
If that same difference applied to 20 images, their combined size would fall by 2160000 bytes. The real decision still depends on each actual image and whether the destination displays WebP, not on extrapolating this example as a guarantee.