Images are almost universally the heaviest assets on a webpage. Google's Core Web Vitals documentation notes that the Largest Contentful Paint element is an image on roughly 70% of all pages. For most sites, that makes an unoptimized image the single factor dragging down search rankings, LCP scores, and conversion rates at the same time.

The fix isn't complicated in theory. In practice, the challenge is scale: hundreds of product photos, thousands of blog images, an entire legacy media library nobody has touched in years. Compressing images one by one isn't a strategy. This guide covers how to do it in bulk, with format and quality decisions applied consistently, so no image needs revisiting twice.

Key Takeaways

• The LCP element is an image on roughly 70% of web pages, so bulk image optimization is the single highest-leverage performance improvement most sites can make.
• Lossy compression is right for photos. Lossless is right for logos, icons, and screenshots. The wrong choice produces either bloated files or degraded visuals.
• WebP is the default format for web images in 2026, with over 97% global browser support.
• The quality sweet spot for photographic WebP is 75 to 82. Higher values add weight without adding visible detail.
• Resize images to their actual display dimensions before compressing. This step alone can cut file sizes by 60% or more.
• Alt text is the half of image optimization most teams forget. Compression handles load speed; alt text handles SEO discoverability and WCAG compliance.

Before You Compress Anything: Lossy vs Lossless

Every compression method falls into one of two categories. Choosing the wrong one for a given image type produces results that are either too large or visibly degraded. This decision happens before you pick a tool.

Lossy compression

Lossy compression permanently removes image data the human visual system is unlikely to notice. At the right settings, a compressed photo looks identical to the original while being 60 to 80% smaller. It works because human vision is highly sensitive to brightness but much less sensitive to subtle color shifts, so lossy algorithms compress color channels more aggressively. At quality 80, the losses are genuinely imperceptible at normal viewing distance. Best for photographs, product images, hero banners, and any image with smooth color gradients.

Lossless compression

Lossless compression reduces file size by encoding data more efficiently, without discarding a single pixel. On photographs it typically achieves 10 to 30% reduction. On logos, icons, and screenshots with large flat-color regions, it can reach 50 to 70% or more, because uniform areas encode very compactly. Best for logos, icons, diagrams, UI screenshots, and any image containing readable text. PNG is the dominant lossless format on the web.

Quick reference by image type

Format Guide: What JPEG, PNG, and WebP Are Built For

Format selection and compression type are two separate decisions. You can apply lossless compression to a PNG and lossy compression to a WebP. Understanding what each format is designed for prevents the most common file-size mistakes.

JPEG

Built for photographic content with smooth tonal gradients. At quality 75 to 85 it's small and clean. Below 70, blocky "mosquito noise" appears around high-contrast edges, especially around embedded text. JPEG can't do transparency or clean sharp geometric edges — never use it for logos, screenshots, or images containing readable text.

PNG

The correct choice when an image needs transparency, sharp edges, or pixel accuracy. The problem is photographic content: a product photo saved as PNG can be 5 to 10 times larger than the same image as WebP. Exporting photos from design tools that default to PNG is one of the most common causes of bloated media libraries.

WebP

WebP supports both lossy and lossless compression, includes transparency, and produces files 25 to 34% smaller than comparable JPEG at equivalent quality. Lossless WebP is roughly 26% smaller than PNG on average. With browser support now exceeding 97% globally, there's no practical reason to default to JPEG or PNG for new web images. The only exception is legacy software pipelines or email clients that don't render WebP.

AVIF

AVIF achieves even smaller files than WebP at equivalent quality, but encoding is slower and tooling less mature. Worth testing on hero images and large product photos if your workflow supports it. AltText.ai supports AVIF and SVG natively, which matters if your library already includes vector or next-generation assets.

Quality Settings: What the Number Actually Controls

The quality slider runs 0 to 100, but it doesn't represent a percentage of the original quality. It controls how aggressively the algorithm discards data, and the relationship is nonlinear. The visible difference between quality 95 and 85 is nearly zero; the file-size difference is substantial.

The practical target for most web photography is quality 78 to 82. A quality-80 file is typically 40 to 50% smaller than a quality-95 file, and they look identical on screen. For WebP, the algorithm is more efficient: quality 75 typically matches JPEG quality 85 in perceived sharpness at a smaller size. Start at 78 to 80 for photographic content, and you can often push to 70 before noticing degradation.

The Bulk Optimization Workflow, Step by Step

Step 1: Audit before you compress

Identify what actually needs fixing. Three categories matter. Oversized images — a 5,000-pixel-wide image displaying at 800 pixels is a resize problem, not a compression problem. Wrong-format images — product photos saved as PNG, logos saved as JPEG. And missing or empty alt text, which is a different kind of performance problem but belongs in the same audit. Images without accurate alt text are invisible to Google Image Search and fail WCAG 2.2 SC 1.1.1. If you're auditing for optimization, audit alt text at the same time — AltText.ai's bulk generation processes entire libraries automatically, so the alt text gap closes in the same workflow window as the compression work.

Step 2: Resize to display dimensions first

Compression reduces bits per pixel. Resizing reduces pixel count. You need both, and resizing comes first. A 6,000 × 4,000 photo at quality 80 is still large if it renders at 1,200 × 800 on your site. Practical targets: standard content images 1,200 to 1,600 pixels wide; retina-optimized up to 2,000 pixels; thumbnails 400 to 600 pixels. Don't upload images wider than 2,000 pixels for standard web content, regardless of source resolution.

Step 3: Select format and compression per image type

Apply the decisions from earlier. Photographs go to lossy WebP at quality 78 to 82. Logos and transparency-dependent graphics stay lossless PNG, or convert to lossless WebP for the size benefit. Screenshots and UI images with text use lossless PNG or WebP. The value of batch processing is consistency: define these settings once and apply them uniformly rather than deciding image by image.

Step 4: Run the batch and spot-check

After the batch runs, open 10 to 15 representative images at 100% zoom and compare against originals. Look at fine textures, sharp edges, embedded text, and smooth gradients. If you see blocking artifacts or color banding, raise quality by 5 points and re-run. At quality 78 to 80 on photographic content, most spot checks show no visible difference.

Step 5: Validate with a performance tool

Publish and test representative pages in Google PageSpeed Insights or WebPageTest. Check whether "Serve images in next-gen formats" and "Efficiently encode images" have cleared. A properly executed bulk pass eliminates both on most pages.

The Mistakes That Silently Undo Optimization Work

Compressing already-compressed files. Running a second round of lossy compression on an already-lossy file compounds the quality loss. Always compress from the highest-quality original — the camera file or the original design export.

Applying one rule to every image type. A media library mixes product photography, brand logos, lifestyle editorial, UI icons, and banners. A single lossy pass across all of them mangles the non-photographic assets. A logo at JPEG quality 75 looks noticeably worse than the same logo as a lossless PNG.

Uploading without resizing. Compression can't fix a 6,000-pixel image used at 800 pixels wide. The unneeded pixels still exist. Skipping the resize step is the most common reason sites see small LCP improvements from compression but still fail Core Web Vitals.

Falling behind on alt text. You can compress every image perfectly and still rank poorly in Google Images if alt attributes are empty, generic, or absent. Compression and accessibility are two sides of the same coin. The WordPress plugin generates alt text automatically at upload, so it doesn't become a separate task that falls through the cracks.

How Bulk Optimization Affects Search Rankings

LCP is a direct ranking factor. LCP measures how quickly the main visible content loads, and Google uses it as a Search ranking signal. Since the LCP element is an image on roughly 70% of pages, image optimization has a direct, measurable effect on visibility — not just speed.

Google Images sends real traffic. Google's image SEO documentation recommends serving images in modern formats, at appropriate sizes, with descriptive alt text. Sites meeting all three appear in Image Search results. For e-commerce, food, travel, and design content, Image Search is a significant organic channel that compression alone doesn't unlock — alt text does.

Crawl budget at scale. Googlebot allocates a finite crawl budget per domain per cycle. Heavy, slow-loading assets consume more budget per visit. Lighter images let Googlebot index more pages in the same window — a meaningful difference for large e-commerce stores and content-heavy publications.

The Part Most Optimization Guides Stop Before

File size and format are what every guide covers. Here's what most don't reach: alt text, and why it belongs in the same workflow. WCAG 2.2 SC 1.1.1 requires every non-decorative image to have a text alternative that conveys the same meaning. It's a Level A requirement, so it applies at every conformance level. Under the ADA, Section 508, and the EU Accessibility Act, missing alt text is one of the most commonly cited violations. Missing alt text is both a legal risk and an SEO gap.

The challenge at scale mirrors the compression challenge. A site with 5,000 product images can't manually write accurate descriptions for each. AltText.ai's bulk generation uses computer vision to describe what's actually in each image rather than repeating the filename, and combined with SEO keyword integration it closes the discoverability gap that compression work leaves open.

Frequently Asked Questions

Does batch compression produce different results than compressing images individually?

No. The algorithm and quality settings are identical either way. The advantage of batch processing is consistency — every image gets the same settings, rather than varying based on whoever handled it manually.

What quality setting should I use for product photos?

WebP quality 78 to 82, or JPEG quality 80 to 85 where WebP conversion isn't part of the workflow yet. This preserves color accuracy, fabric texture, and fine detail while delivering much smaller files. For category-page thumbnails, quality 70 to 75 is appropriate.

Should I convert existing PNG product images to WebP?

For photographic PNGs, yes — the size improvement is significant and the visual result is identical. For logos and icons correctly saved as PNG with transparency, converting to lossless WebP delivers additional size reduction without data loss. For UI screenshots with text, lossless WebP and PNG both work.

How should I approach a legacy media library that's never been optimized?

Start with the highest-traffic pages: homepage, top landing pages, and highest-revenue product or category pages. Optimize those first and measure LCP before moving to the rest of the library.

Does alt text affect where my images rank in Google Image Search?

Yes. Alt text is the primary text signal Google uses to understand what an image depicts and which queries it should appear for. Compressed images with no alt text are invisible to Image Search; compressed images with accurate, descriptive alt text are indexable.

Related

• Best Alt Text Generators Compared
• Why Keyword Stuffing Alt Text Still Fails
• WCAG Alt Text Guide
• WordPress Alt Text Integration Guide