UtiliKit
Back to blog

JPG to WebP for Site Speed: Real Size Savings on 10 Photos

Published: November 11, 2026 - 6 min read

Converting JPG to WebP for website speed saved 33% on average in my test of 10 photo-style images, when both formats were encoded fairly at the same size and quality setting. Straight from the original files the saving looked like 62%, but most of that came from the originals being saved at a very high JPEG quality, not from WebP being magic. The honest range per image was 16% to 54%, and the image that saved least is the kind you'd least expect.

Here's what I tested, the exact numbers, and the settings I'd use on a real site.

The 10 test images

I used 10 photorealistic JPEGs: seven interior scenes and three house exteriors, the sort of large hero and gallery images that make up most of the weight on a typical small business site. Nine were 1920 px wide (1080, 1440 or 1920 px tall), one was 1448 × 1086. Original sizes ran from 403 KB to 1.27 MB, 5.95 MB in total.

Each one went through two tests. First, a straight conversion at full size to WebP quality 80. Second, a fair comparison: resize to 1600 px wide (a realistic width for a full-width image on a laptop), then save both as JPEG quality 80 (optimised, progressive) and as WebP quality 80. Encoding was done with libwebp and libjpeg at their standard settings, which is the same codec family browser-based converters use. Expect your own numbers to differ by a few percent.

Results, image by image

All sizes in KB, both formats at 1600 px wide (the 1448 px image stayed at its own width):

ImageOriginal JPGJPG q80WebP q80WebP saving vs JPG q80
Interior 1, 1920 × 108052821816027%
Exterior 1, 1920 × 192088235623135%
Interior 2, 1920 × 14404041617454%
Interior 3, 1920 × 108046519512635%
Exterior 2, 1920 × 19201,26851042816%
Interior 4, 1448 × 108642824218424%
Interior 5, 1920 × 108042218512035%
Exterior 3, 1920 × 108043317812132%
Interior 6, 1920 × 192066828415047%
Interior 7, 1920 × 19204521949451%
Total5,9492,5251,68733%

For the straight full-size conversion (no resize, WebP q80), the 10 files went from 5,949 KB to 2,258 KB.

That's 62%.

It's the kind of figure most "convert to WebP" pages quote. Real, but misleading.

Why the big number is mostly not WebP

Look at the second column. Simply re-saving the same images as JPEG quality 80 at 1600 px cut the total from 5.95 MB to 2.53 MB, a 58% drop, and I didn't touch the format at all. The originals were exported at a high quality setting (likely 90 or above) and at a width bigger than most layouts ever display.

So if your site is full of camera-quality JPEGs, the first win is resizing and sensible quality. WebP then takes another third off on top. Both are worth having, but it helps to know which one is doing the heavy lifting, because resizing works for every browser, every email client and every social platform, while WebP doesn't (more on that below).

The Image Resizer handles the first step, and the JPEG to WebP converter handles the second, both in the browser without uploading the files anywhere.

Which images gain most (and least)

The two best results, 54% and 51%, were interiors with large smooth areas: plain walls, soft gradients from window light, a pale floor. WebP's block prediction handles smooth tonal changes far better than JPEG's 8 × 8 blocks, so there's simply less to store.

The worst result was an exterior with dense trees, grass and gravel.

Only 16% smaller.

Fine, random detail is expensive in every format, and there's little structure for WebP to exploit. Busy textures like foliage, crowds, fabric close-ups or sensor noise from a phone photo in low light will save less than the average.

A quick rule from this test: clean product shots and interiors often save 40 to 50%, busy outdoor scenes 15 to 25%.

The quality number trap

Quality 80 in WebP is not the same as quality 80 in JPEG. The scales aren't calibrated against each other, and that catches people out.

When I pushed WebP to quality 90 on the same 10 images, the total came to 2,965 KB, which is bigger than JPEG at quality 80. Anyone who converts with a "high quality, just to be safe" setting can end up with WebP files larger than the JPEGs they replaced, and then conclude that WebP doesn't work. Going the other way, WebP at quality 75 came to 1,342 KB, 47% below the JPEG q80 total, and on screen I couldn't spot a difference at 100% zoom on most images. Grain in the darkest corners of two interiors got slightly smoother, which you'd only notice side by side.

The converter's own advice of 80 to 85% matches what I saw. Below 70, smooth walls start to show faint banding.

What I'd actually do on a real site

For a typical small site with a dozen pages of photos, this is the order I'd follow:

  1. Resize every image to the largest width it's displayed at, roughly 1600 px for full-width images and 800 to 1000 px for content images. Use 2× only where people zoom, such as product galleries.
  2. Convert to WebP at quality 80. Drop to 75 for big backgrounds that sit behind text, where nobody inspects detail.
  3. Keep JPEG copies for email newsletters and for the social sharing image (og:image), because some email clients and link preview systems still handle WebP poorly.
  4. Check a few pages before and after with the SEO Analyzer and your browser's network tab, and look at the total image weight of the page rather than one file.

Every current major browser displays WebP, so for images on your own pages you don't need a JPEG fallback any more unless you support very old devices.

Is a 33% saving worth the effort? On a gallery page carrying 3 MB of images, it's roughly 1 MB less to download on every visit, and the largest image on the page (usually the hero) appears sooner. That's the image that decides your Largest Contentful Paint, which is one of the speed numbers search engines look at. It won't move rankings on its own, but slow hero images are one of the most common reasons a small site fails its speed check.

When not to bother

If your images are already small (under about 100 KB each) and correctly sized, converting them will save a few kilobytes per page. Not worth an afternoon. Skip it. Same for logos and icons: those should be SVG or PNG anyway, and there's a separate guide on JPG vs PNG vs WebP per use case.

And if you need a specific file size rather than a quality setting, for example a 200 KB limit from a site builder, the Compress to a File Size tool targets the encoded size directly.

To try it on your own photos, start with one heavy page: resize, convert a handful of images with the JPEG to WebP converter, and compare the totals. The free version covers that. If you've got hundreds of product images to do in one go, batch conversion is part of UtiliKit Premium.

Frequently asked questions

How much smaller is WebP than JPG?

In my test of 10 photos at the same width and quality 80, WebP files were 33% smaller on average, with a range of 16% to 54%. Smooth images such as interiors and product shots save the most, busy outdoor scenes the least.

What WebP quality should I use for a website?

Quality 80 is a good default for photos. Quality 75 works well for large background images. Avoid 90 and above, because WebP files at that setting can end up larger than the original JPEGs.

Do I still need JPG versions of my images?

For images on your own web pages, current browsers all display WebP. Keep JPG copies for email newsletters and for the social sharing image, where support is less consistent.

Are my photos uploaded when I convert them?

No. UtiliKit's converter runs in your browser, so the photos are processed on your own device and are not uploaded to a server.

More guides