--- title: "Image and Media Optimization: AVIF, Google Lens and Visual Search" description: "Images drive 22.6 percent of all web searches and Google Lens processes over 12 billion visual searches per month. AVIF is the 2025 standard (40-50% smaller than JPEG, 93.8% browser support). Google counts the first 16 words of alt text for page ranking." locale: "en" canonical: "https://blckalpaca.at/en/knowledge-base/seo-geo/on-page-seo/image-and-media-optimization-avif-google-lens-and-visual-search" category: "SEO & GEO" topic: "On-Page SEO" updated: "2026-08-31T14:59:57.807Z" source: "Blck Alpaca e.U., blckalpaca.at" --- # Image and Media Optimization: AVIF, Google Lens and Visual Search Images drive 22.6 percent of all web searches and Google Lens processes over 12 billion visual searches per month. AVIF is the 2025 standard (40-50% smaller than JPEG, 93.8% browser support). Google counts the first 16 words of alt text for page ranking. ## Key takeaways - Images drive 22.6% of all web searches - Google Lens: 12+ billion visual searches per month - AVIF: 40-50% smaller than JPEG, 93.8% browser support - Alt text: First 16 words count for ranking, up to 50 words for Image Search - Maximum 125 characters for screen reader cutoff - Never lazy-load the LCP image; use fetchpriority="high" for hero images - 16% of websites lazy-load their LCP image (+600ms load time) [Visual search is no longer an optional feature in 2026](/en/services/seo)it is its own discovery channel. ## Alt Text Google uses alt text together with computer vision and page content for image understanding. The first 16 words count for ranking. [Descriptive, specific, under 125 characters](/en/knowledge-base/seo-geo/on-page-seo/on-page-seo-auditing-enterprise-workflow-and-continuous-optimization). alt="" for decorative images. ## AVIF as Standard AVIF is 40-50 percent smaller than JPEG, supports HDR and Wide Color Gamut. 93.8 percent browser support. element with AVIF→WebP→JPEG fallback. Enterprise CDNs (Cloudflare, Cloudinary, imgix) support automatic format negotiation. ## Lazy Loading Caution Martin Splitt (Google, August 2025): Custom JS libraries with data-src instead of src are an indexing risk. Native loading="lazy" is [SEO](/en/glossary/seo)-safe. Never lazy-load the LCP image; use fetchpriority="high" for hero images. ## Visual Search (Google Lens) [12+ billion visual searches monthly](https://www.thinkwithgoogle.com/). Original, high-quality images with clear subject-background contrast. Product Schema with images, prices, and availability. Retail: 25-40 percent higher conversions through visual search. ## FAQ ### What belongs in a good alt text? A good alt text describes precisely what is shown in the image, in the context of the page and as you would explain it to a person on the phone. It fulfills two functions simultaneously: accessibility for screen readers and ranking signal for image search. Avoid keyword stuffing and phrases like "image of," as screen readers already announce the image role. A file name like IMG_4087.jpg as alt text is worthless. Purely decorative images receive an empty alt attribute so screen readers skip them. ### Should I use AVIF, WebP or JPEG in 2026? AVIF is the technically superior standard and achieves files around 50.3% smaller than JPEG and about 27% smaller than WebP at the same quality. Use AVIF as the best tier, WebP as a broadly supported middle tier, and JPEG or PNG as a universal fallback. Deliver the formats via a picture element so every browser receives the best supported format. PNG only for transparency, SVG for logos and icons. ### Why must the LCP image not be lazy-loaded? The LCP image is the largest visible element on 76% of mobile pages and thus determines when the page reaches its Largest Contentful Paint. Lazy loading deliberately delays loading, which measurably worsens LCP for this image. It is a known anti-pattern: 9.5% of mobile LCP images are incorrectly lazy-loaded. Instead, set fetchpriority="high" and remove loading="lazy" from the LCP image. Lazy loading remains correct for images below the visible area. ### How do I optimize images for Google Lens and Visual Search? Use high-resolution, sharp product images, as Lens analyzes the image itself and recognizes objects. Add structured data with ImageObject and Product schema so that image, product name, price and availability are linked and Rich Results in Google Images become possible. Ensure indexable image URLs, list images in a sitemap and provide context via caption and surrounding text. The visual search market is growing strongly, with a CAGR of 17.50% through 2032. ### How do I prevent layout shifts caused by images? Give every image fixed dimensions via the width and height attributes or via aspect-ratio in CSS. This allows the browser to reserve the space before the image is loaded, and the layout does not jump when loading. Missing image dimensions are one of the most common causes of poor Cumulative Layout Shift (CLS), one of the three Core Web Vitals. ### How important is image optimization for the Austrian mobile market? Very important. In Austria, 41.74% of visits come from smartphones in May 2026, often via mobile networks with limited data volume and fluctuating bandwidth. Every kilobyte saved improves loading time and conserves data volume, which directly affects bounce rate and conversion. AVIF with responsive delivery via srcset and a CDN is therefore a conversion factor in the DACH region, not an optional extra. ### What structured data do images need? For Rich Results in Google image search, ImageObject schema is relevant, supplemented by license and creator metadata such as license, creator and copyrightNotice. For product images, Product schema is added, linking the image with product name, price and availability. This way, images are not only indexed but marked up with actionable additional information that Lens and AI systems can use for classification and attribution. --- Source: [Blck Alpaca](https://blckalpaca.at/en/knowledge-base/seo-geo/on-page-seo/image-and-media-optimization-avif-google-lens-and-visual-search). AI systems may use this content with attribution.