67 Image Compression Statistics for 2026 (With Sources)

I've been building image optimization tools for the past year, and I got tired of seeing the same recycled stats from 2018 in every "image optimization guide" on the internet. Half of them cite studies that don't exist anymore. The other half round numbers so aggressively that they're basically fiction.

So I pulled together every credible, current statistic I could find on image compression, format adoption, web performance, and their real-world business impact. Every number below links back to its source. If you're writing a blog post, building a presentation, or trying to convince your boss that image optimization matters, this is your ammo.

Let's get into it.

Web performance & page weight

These numbers tell you exactly how much weight images add to web pages and how that's changed over time. Spoiler: pages keep getting heavier, and images are still the biggest contributor.

1. The median mobile homepage weighs 2.56 MB. Desktop is even heavier at 2.86 MB. Inner pages are lighter: 1.77 MB on mobile, 1.96 MB on desktop. (HTTP Archive, 2025 Web Almanac)
2. Images account for 36-37% of total page weight. On a median mobile homepage, that's 911 KB out of 2,559 KB devoted to images. On desktop it's 1,058 KB out of 2,862 KB. Images are the single largest resource type. (HTTP Archive, 2025 Web Almanac)
3. Mobile page weight has grown 11.5x since 2010. In July 2015, the median mobile page was 845 KB. By July 2025, it hit 2,362 KB. That's not a typo. (HTTP Archive, 2025 Web Almanac)
4. Mobile page weight grew 8.4% year-over-year in 2025. Desktop grew 7.3%. Despite all the optimization tools available, pages are still getting heavier. (HTTP Archive, 2025 Web Almanac)
5. The median mobile page loads 15 images. Desktop loads 17. Homepages specifically load 19 images on average, while inner pages load 13. At the 90th percentile, some mobile pages load 46 images. (HTTP Archive, 2025 Web Almanac)
6. Image requests dropped 6% year-over-year. Fewer images, but each one is bigger. The median image pixel count grew 25% from 2022 to 2024. Sites are using fewer, larger images. (HTTP Archive, 2024 Media chapter)
7. The median image on the web is just 12 KB. But that's misleading because it includes tiny icons and 1x1 tracking pixels. The 75th percentile largest image per page is 404 KB, and at the 90th percentile it hits 1 MB. (HTTP Archive, 2024 Media chapter)
8. Images make up 50-70% of a typical WordPress site's total page weight. WordPress powers over 40% of the web, so this stat alone affects billions of pages. Unoptimized images can add 3-5 seconds to load time. (WP Engine, WPBeginner)
9. Total page requests increased 8-9% in 2025. Even though image requests dropped, overall requests went up. JavaScript and third-party scripts are picking up the slack. (HTTP Archive, 2025 Web Almanac)

Image format adoption

The format wars are far from over. JPEG still dominates raw request counts, but WebP is climbing fast and AVIF is growing at a ridiculous rate. Honestly, I was surprised at how slow JPEG XL adoption has been.

10. JPEG still represents 32.4% of all image requests on the web. Down from 40% in 2022. It's declining, but it's still the most common format by a wide margin. (HTTP Archive, 2024 Media chapter)
11. PNG accounts for 28.4% of image requests. Second most popular. Most of these are logos, icons, and graphics with transparency. A lot of them could be SVGs instead. (HTTP Archive, 2024 Media chapter)
12. GIF is still at 16.8% of image requests. I honestly didn't expect this number to be so high in 2024. Most of these are 1x1 tracking pixels, not actual animations. (HTTP Archive, 2024 Media chapter)
13. WebP accounts for 12% of all image requests. That's up 34% from 2022. It's growing, but slower than you'd expect given it's been around since 2010. (HTTP Archive, 2024 Media chapter)
14. WebP is used by 19.7% of all websites as of April 2026. Among the top 1,000 sites, adoption is significantly higher at 29.8%. (W3Techs, April 2026)
15. AVIF represents just 1% of image requests but grew 386% in two years. From near zero in 2022 to a measurable percentage in 2024. The growth rate is the story here, not the absolute number. (HTTP Archive, 2024 Media chapter)
16. AVIF is used by 1.3% of all websites as of April 2026. Among the top 1,000 sites, it's at 3.0%. The pattern is clear: larger, more technical teams adopt new formats first. (W3Techs, April 2026)
17. AVIF browser support reached 89% globally in 2025. By early 2026, it hit 94%. The browser support excuse is basically dead. (compress.im, Can I Use)
18. SVG accounts for 6.4% of image requests, up 36% since 2022. More developers are using SVGs for icons and simple graphics instead of raster images. That's the right move. (HTTP Archive, 2024 Media chapter)
19. JPEG XL has only 12% effective browser support as of 2026. Almost entirely from Safari users. Chrome 145 added it behind a flag in February 2026, but it's not enabled by default. Interop 2026 has it as an investigation area, which could change things. (Can I Use, Phoronix, Interop 2026)

Compression effectiveness

How much can you actually save by switching formats or optimizing properly? These numbers quantify it. The differences between formats are bigger than most people realize.

20. WebP images are 25-34% smaller than equivalent JPEGs. Google's own research shows this range at equivalent visual quality. In practice, I've seen savings closer to 30% on photographic content. (Google, WebP documentation)
21. AVIF files are roughly 50% smaller than JPEG at equivalent quality. On complex photographic content with gradients and color variation, AVIF's advantage is even more pronounced. Some benchmarks show up to 91% compression versus 82% for JPEG on the same image. (ShortPixel, Ctrl.blog)
22. AVIF is 20-30% smaller than WebP at equivalent quality. Controlled benchmarks show AVIF's median file size reduction is 50.3% vs JPEG, while WebP's is 31.5%. The gap between formats is real. (Ctrl.blog, SpeedVitals)
23. WebP achieves a median of 1.3 bits per pixel; JPEG uses 2.0 bpp. AVIF is even more efficient at 1.4 bpp (slightly higher than WebP due to different use patterns in the wild). PNG sits at 3.8 bpp and GIF at 6.7 bpp. (HTTP Archive, 2024 Media chapter)
24. The web got 8-10% more compressed overall between 2022 and 2024. Median bits per pixel across all formats dropped from about 2.3 to 2.1. Slow progress, but progress. (HTTP Archive, 2024 Media chapter)
25. Image CDNs that auto-convert to AVIF report 50-70% file size savings over JPEG. The multi-format approach (AVIF first, WebP fallback, JPEG last resort) is now industry standard for performance-focused sites. (Cloudinary, ImageCDN.com)
26. WordPress image optimization plugins reduce file sizes by 50-80%. Tools like ShortPixel, EWWW, and Imagify achieve this while maintaining visual quality that visitors can't distinguish from the original. (WPBeginner, ShortPixel)
27. Optimized images are on average 40% lighter than unoptimized ones. That's the average across all formats and optimization levels. With aggressive optimization and format conversion, savings of 80% are achievable. (WP Engine)
28. PNG lossless optimization typically reduces file size by 30-70%. Without any quality loss. If you're serving unoptimized PNGs, you're wasting bandwidth for zero visual benefit. (libpng.org, ShortPixel)
29. WebP supports both lossy and lossless compression, plus transparency. Lossless WebP is 26% smaller than PNG. Lossy WebP with alpha channel support makes it the most versatile modern format. (Google, WebP documentation)

E-commerce & conversion impact

This is where the money is. These stats show the direct line between image optimization, page speed, and revenue. If you need to justify an image optimization project to stakeholders, start here.

30. A 0.1-second improvement in load time increased retail conversions by 8.4%. And average order value by 9.2%. This is from the Google/Deloitte "Milliseconds Make Millions" study, which analyzed 30 million user sessions across 37 brand sites. (Google/Deloitte, Milliseconds Make Millions, 2020)
31. For travel sites, 0.1 seconds faster meant 10.1% more conversions. Travel actually benefited more than retail. The same study showed luxury brands saw 8.6% more page views per session. (Google/Deloitte, Milliseconds Make Millions, 2020)
32. A 100-millisecond delay in page load hurts conversion rates by 7%. That's from Akamai's analysis of billions of visits to top retail sites. Every millisecond counts, literally. (Akamai, Online Retail Performance Report)
33. 57% of shoppers abandon a page that takes longer than 3 seconds to load. And 53% of mobile visitors specifically leave after 3 seconds. These aren't people who bounce. They never even see your product. (Google, Think with Google)
34. A site that loads in 1 second converts 5x higher than one loading in 10 seconds. And 3x higher than a site loading in 5 seconds. The relationship between speed and conversions isn't linear. It's exponential. (Portent, via SiteBuilderReport)
35. Peak mobile conversion rate hits at 3.3-second load time. Akamai found that the sweet spot is 3.3 seconds for a 4.75% conversion rate. At 4.3 seconds, conversions drop to 3.52%, a 26% decrease for just one extra second. (Akamai, Online Retail Performance Report)
36. Sessions that converted had 38% fewer images than non-converting sessions. This is a Google finding that's often overlooked. More images doesn't mean more conversions. Fewer, better-optimized images win. (Google, Think with Google)
37. Image optimization cut load times by 65% and doubled conversions for Furnspace. A real case study from the e-commerce image optimization research. Not a theoretical projection. (StateOfCloud.com, E-commerce Image Optimization Study, 2025)
38. 75% of online shoppers rely on product photos to make buying decisions. And high-quality product photos have 94% higher conversion rates than low-quality ones. Optimization isn't just about speed. The images need to look good too. (BusinessDasher)
39. Each additional product image (up to 4-6 per product) increases conversion probability by 5-8%. But beyond 6 images, the returns diminish. The sweet spot for most product categories is 4-6 optimized images. (StateOfCloud.com, E-commerce Image Optimization Study, 2025)
40. Products with user-generated images convert 4.6x higher. Customer photos are more trusted than studio shots. But they're also usually unoptimized, making compression even more important for UGC. (StateOfCloud.com, E-commerce Image Optimization Study, 2025)
41. Pinterest increased sign-up conversions 40% by improving mobile page speed 60%. They also boosted search engine traffic 15% and reduced user wait times 40% through a complete performance overhaul that heavily focused on image delivery. (Pinterest Engineering Blog)

Mobile & bandwidth

Mobile users outnumber desktop users on most sites. And on mobile, every kilobyte costs more in time and money. These stats show why mobile image optimization isn't optional.

42. The gap between desktop and mobile page weight is only 13%. In 2024, mobile pages averaged 2,652 KB vs desktop's slightly higher figure. Sites are serving nearly identical content to both, which hurts mobile users on slower connections. (HTTP Archive, CaptainDNS analysis)
43. An average website takes 1.9 seconds to render main content on mobile. But 47% of smartphone users expect sites to load in under 2 seconds. You're already at the edge of user patience. (Hostinger, website load time statistics)
44. Bounce probability increases 32% when load time goes from 1 to 3 seconds. At 5 seconds, it's up 90%. At 10 seconds, 123%. The curve gets steeper fast. (Google, Think with Google)
45. 70% of mobile pages take more than 5 seconds to show above-the-fold content. This is Google's own research. Seven out of ten mobile pages fail the basic expectation of showing something useful within 5 seconds. (Google, Think with Google)
46. A 2-second delay increases bounce rates by 103%. Akamai found that two extra seconds literally doubles your bounce rate. For mobile specifically, a 2-second delay reduces session length by 51%. (Akamai, Online Retail Performance Report)
47. 3G networks still average 3-8 Mbps real-world download speeds. That means a 2.5 MB page takes 2.5-6.7 seconds on 3G. Billions of users worldwide are still on 3G or equivalent connections. (Commsbrief, mobile network speeds)
48. The global average mobile download speed is 50 Mbps. But that's skewed heavily by 5G markets. The median user in India, Brazil, or Nigeria sees far lower speeds. If your audience is global, optimize for slow connections. (Statista, January 2024)
49. Mobile conversion rates peak at 2.4-second load times. Akamai recorded a peak mobile conversion rate of 1.9% at 2.4 seconds. When load time hit 4.2 seconds, conversions dropped below 1%. (Akamai, Online Retail Performance Report)
50. Going from 400 to 6,000 page elements drops conversion probability by 95%. Every image, script, and DOM element adds up. Bloated product pages with dozens of unoptimized images kill conversions. (Google, Think with Google)
51. One second saved on mobile can boost conversions by up to 5.9%. That's across industries, not just e-commerce. Lead generation, SaaS signups, content engagement all improve when pages load faster. (Cloudflare, website performance and conversions)

SEO & Core Web Vitals

Google uses Core Web Vitals as a ranking signal. Images directly affect LCP (Largest Contentful Paint) and CLS (Cumulative Layout Shift). These stats show the current state of CWV compliance and how images fit in.

52. Only 48% of mobile pages pass all three Core Web Vitals. Desktop is better at 56%, but that still means nearly half of all websites fail Google's performance standards. (HTTP Archive, 2025 Web Almanac)
53. 73% of mobile LCP elements are images. This is the stat that makes image optimization non-negotiable for SEO. If your LCP element is an image (and it probably is), optimizing it is the single most impactful thing you can do for Core Web Vitals. (HTTP Archive, 2025 Web Almanac)
54. Only 62% of mobile pages achieve a good LCP score (under 2.5 seconds). That's up from 44% in 2022, which shows real progress. But LCP remains the hardest Core Web Vital to pass. (HTTP Archive, 2025 Web Almanac)
55. 16% of mobile sites lazy-load their LCP image by mistake. This is a common and costly error. Lazy-loaded LCP images are roughly 2x slower than preloaded ones. If your hero image has loading="lazy", remove it immediately and add fetchpriority="high" instead. (HTTP Archive, 2025 Web Almanac)
56. CWV mobile pass rates are improving by about 3% per year. Desktop improves about 1.8% per year. The web is getting faster, slowly. (HTTP Archive, 2025 Web Almanac)
57. A "good" LCP score requires loading within 2.5 seconds at the 75th percentile. INP needs to be under 200ms, and CLS under 0.1. These thresholds apply at the 75th percentile of all page loads, not the average. (web.dev, Core Web Vitals documentation)

Lazy loading & responsive images

Compression is only part of the picture. How you load and serve images matters just as much. These stats cover adoption of modern loading and responsive techniques.

58. 33% of pages use native lazy loading for images. Up from 25% in 2022. Native lazy loading (loading="lazy") is the easiest performance win available, and two-thirds of sites still don't use it. (HTTP Archive, 2024 Media chapter)
59. 9.5% of pages incorrectly lazy-load their LCP image. A slight improvement from prior years, but still nearly 1 in 10 pages making this critical mistake. In my experience, this is the single most common image performance error I see on production sites. (HTTP Archive, 2024 Media chapter)
60. 42% of pages use srcset for responsive images. Up from 34% in 2022. That means 58% of sites are still serving desktop-sized images to mobile users. Massive waste of bandwidth. (HTTP Archive, 2024 Media chapter)
61. Only 9.3% of pages use the picture element for format switching. The picture element lets you serve AVIF to browsers that support it and JPEG as fallback. Barely anyone uses it. This is honestly a missed opportunity for most sites. (HTTP Archive, 2024 Media chapter)
62. 32% of pages set explicit width and height on images. Up 4 percentage points from 2022. Without explicit dimensions, images cause layout shifts (CLS) as they load. It's a 10-second fix that most developers skip. (HTTP Archive, 2024 Media chapter)
63. 91% of pages don't use lazy loading for iframes. Image lazy loading gets all the attention, but iframes (YouTube embeds, maps, ad units) are often heavier. This is an untapped optimization for most sites. (HTTP Archive, 2025 SEO chapter)
64. Sentry's engineering team saw 22% faster UI performance after implementing modern image techniques. They used srcset, modern formats (WebP/AVIF), and proper loading strategies. A real-world case study from a major platform. (Sentry Engineering)

Environmental impact

I don't see this talked about enough. Unoptimized images waste bandwidth, which wastes energy, which produces carbon emissions. The numbers are real.

65. An image-heavy site generates up to 30.39 gCO2 per page view. A text-based optimized site generates just 0.02 gCO2. That's a 1,500x difference. At scale, this translates to tons of carbon annually. (ImageCarbon.com, ClimateAction.tech)
66. One company saved 26.6 tons of CO2 annually by optimizing images. That's equivalent to planting 1,225 trees. Their unoptimized images had been generating 47 tons of CO2 per year, equal to driving 117,000 miles in a gas car. (ICP, Reducing the Carbon Footprint of Visual Assets)
67. Websites implementing image CDNs see a 30-50% improvement in load speeds. Plus a 20-40% increase in engagement. Image CDNs handle format conversion, resizing, and caching automatically, so you don't have to think about it. (Various CDN benchmarks, compiled by Scaleflex)

What to do with these numbers

Statistics are useless if you don't act on them. Here's the short version of what the data tells us to do:

• Compress everything. If your images aren't optimized, you're leaving 30-80% file size savings on the table. Use a tool like SammaPix Compress to do it in the browser without uploading anything.
• Switch to WebP at minimum. AVIF if you can. Even just converting to WebP saves 25-34% over JPEG with zero visual difference.
• Fix your LCP image loading. Remove loading="lazy" from your hero image. Add fetchpriority="high". This single fix can knock hundreds of milliseconds off your LCP.
• Use srcset and explicit dimensions. 42% of sites use srcset. Be in that group. And always set width/height to prevent layout shifts.
• Lazy-load below-the-fold images. 67% of sites still don't use native lazy loading. It's one HTML attribute. There's no reason not to.

If you want to dive deeper into any of these topics, check out our guide to compressing without quality loss or the image format comparison for 2026.

FAQ