Efficiently Encode Images: Compress Without Killing Quality

What this means

This audit is about compression efficiency, not format or dimensions. Lighthouse gathers every JPEG and BMP, re-encodes each at compression level 85, and compares against what you served. When the re-compressed copy is at least 4 KiB lighter, the image is flagged. The file simply carries more data than the eye needs at its display size.

Usual culprits are exports saved at quality 95 to 100, raw stock photos, and files heavy with camera metadata (EXIF, GPS, color profiles). As of Lighthouse 13 this audit was folded into the broader "Improve image delivery" insight, so you may see it under that label.

Why it matters

Images are usually the heaviest thing on a page, so over-encoded files hurt twice.

Speed. The largest image very often defines your Largest Contentful Paint. A hero photo carrying excess bytes delays the moment Google measures as "loaded." Trim it and LCP improves directly, no code changes needed.

Data and reach. Bloated images waste visitor bandwidth and hosting transfer, and punish people on slow or capped connections most. Efficient encoding is a high-leverage, low-effort win that feeds your wider image SEO strategy.

How it gets flagged

Lighthouse, and tools built on it like PageSpeed Insights and GTmetrix, runs the check automatically:

• Collects every JPEG and BMP on the page.
• Re-encodes each at compression level 85.
• Compares the re-encoded size to what you served.
• Lists any image where the saving is 4 KiB or more, with its current size and estimated saving.

PNG, WebP, AVIF, GIF, and SVG are not assessed here; their savings surface under other audits. The table is your work list: sort by saving, start at the top.

How to fix it

1. Compress at a sensible quality

Re-export JPEGs at quality 80 to 85, not 95 to 100. That is the sweet spot Lighthouse itself uses, almost always invisible while cutting size sharply. Squoosh, ImageOptim, and TinyPNG let you preview first.

2. Strip metadata

Remove EXIF data, GPS coordinates, thumbnails, and bulky color profiles. Most compression tools do this automatically; jpegoptim --strip-all handles it on the command line.

3. Use modern formats

WebP or AVIF compress far better than JPEG at the same quality, clearing this audit and the related next-gen formats check. Convert once, keep a JPEG fallback for old browsers.

4. Automate it

On WordPress, plugins like Imagify, ShortPixel, Smush, or EWWW compress every upload, often converting to WebP too. An image CDN (Cloudinary, imgix, Cloudflare Images) compresses, resizes, and reformats on the fly. Set it once and the audit stays green.

Related image audits

Three image audits sound alike but fix different problems; one image can fail several at once:

• Efficiently encode images (this one): right format and size, but compressed too lightly. Fix by lowering quality and stripping metadata.
• Serve images in next-gen formats: the format is dated. A well-compressed JPEG can still fail because WebP or AVIF would be smaller. Fix by converting.
• Properly size images: pixel dimensions exceed the display space, so the browser shrinks it. Fix by serving responsive sizes.

Compressed well (encode), modern format (next-gen), right dimensions (size). Good delivery answers yes to all three.

False positives and edge cases

The check is conservative, but a few flags deserve a second look. Savings are estimated at a fixed quality 85; if an image needs higher fidelity (fine gradients, product photography), that may visibly degrade it, so judge by eye. Third-party and ad-network images may be flagged even though you cannot re-encode them. Fix the heavy hitters first; a fast, sharp page beats a perfect 100.

FAQ