Image SEO is not a single compression score. It is the combined result of a useful image, an appropriately small file, correct HTML, and fast, stable delivery.
1. Start with display dimensions
Export close to the largest size the layout actually renders. A 4000px camera photo shown in a 900px article column transfers and decodes unnecessary data. Keep a higher-resolution source outside the website, then create delivery-sized copies.
2. Pick the format by content
JPEG remains practical for compatible photographic delivery. WebP is a dependable modern default for photos and transparency. AVIF can achieve smaller results but is more expensive to encode. PNG is best reserved for lossless graphics or cases where its exact behavior is needed.
3. Compress with a visual check
For photos, try 75–85% quality and inspect faces, text, gradients, and sharp edges. The right value depends on the image. A tool should show the output rather than promise “no quality loss” for every lossy operation.
4. Prevent layout shift
Include width and height attributes or an aspect-ratio container so the browser reserves space before the image arrives. This supports visual stability and avoids content jumping during load.
5. Write useful context
Use concise descriptive filenames and alt text that explains the image's purpose. Decorative images can use empty alt text. Do not stuff keywords into captions, filenames, or alternative text.
6. Deliver responsibly
Use responsive srcset candidates, lazy-load below-the-fold media, preload only genuinely important hero media, and measure real-user Core Web Vitals after deployment.
Process it privately in your browser.