Image SEO is the practice of optimising the images on your pages so they load fast, get indexed, and improve both your rankings and your visibility in Google Images. Done well, it means serving the right format at the right size with markup that search engines and screen readers can actually read. Done badly, images become the heaviest, slowest, and most invisible part of your site.
Most teams treat images as decoration and never look at them again after upload. That is a mistake, because images are usually the biggest single lever you have over page weight and one of the most neglected. This guide covers formats, compression, on-page signals, indexing, and a checklist you can run against any page.
Why Image SEO Still Moves the Needle
Images are almost never the reason a page ranks, but they are frequently the reason it does not. The HTTP Archive’s Web Almanac consistently finds images to be one of the heaviest resource types on the median page, and unoptimised images are the most common cause of a poor Largest Contentful Paint score. Since Core Web Vitals is a confirmed ranking signal, that connection is direct and measurable.
There is a second, more obvious benefit. Google Images is a genuine discovery surface, and for recipe sites, ecommerce, travel, and any visual niche it can send meaningful traffic. Google’s own guidance is explicit that image results and Discover reward images that are high quality, correctly marked up, and served on fast pages.
So image SEO works on two fronts at once. It protects your existing rankings by keeping pages fast, and it opens a secondary traffic channel through image search. Both come from the same handful of technical decisions, which is what makes this one of the highest-return areas in a technical SEO audit.
The mistake I see most often is treating these as separate projects. They are not. Fix the format and the file size, and you improve Core Web Vitals and image-search eligibility in the same pass.
Choose the Right Format: AVIF, WebP and the Fallbacks
Format choice is the highest-leverage decision in image optimisation, and most sites are still shipping the wrong one. For photographs and complex imagery, modern formats crush the older ones on file size at equivalent quality.
Google’s WebP documentation reports that lossless WebP images are around 26 percent smaller than PNGs, and lossy WebP images are 25 to 34 percent smaller than comparable JPEGs at equivalent quality. AVIF, the newer of the two next-gen formats, typically compresses even harder, often halving JPEG file sizes. Both are now supported across every major browser.
Here is the decision framework I use:
- Photographs and rich imagery → AVIF first, WebP as the widely supported default, JPEG as the legacy fallback.
- Logos, icons, and line art → SVG, which is resolution-independent and usually a fraction of a kilobyte.
- Images needing lossless transparency → WebP, or PNG only where you must.
- Animations → a video file (MP4/WebM), never an animated GIF, which is grotesquely inefficient.
The clean way to ship modern formats with a safe fallback is the <picture> element, which lets the browser pick the first source it supports. If that is too much markup to maintain by hand, an image CDN that negotiates format and size automatically per request is worth the money on any image-heavy site.
Do not skip the fallback. Serving AVIF-only to save effort will break images for a slice of your audience, and a broken image is worse for both users and crawlers than a slightly larger one.
Image SEO and Core Web Vitals
This is where image SEO and page speed become the same discipline. On most content pages the Largest Contentful Paint element is an image, usually the hero or a featured graphic near the top of the viewport. That means the single image at the top of your page often decides whether you pass the LCP threshold of 2.5 seconds that Google uses for a “good” Core Web Vitals score.
Three things dominate that outcome. First, file size: compress the image and serve a modern format so fewer bytes travel down the wire. Second, dimensions: never ship a 3000-pixel-wide image into a 800-pixel container, and use responsive srcset so each device downloads an appropriately sized file. Third, loading priority: the LCP image should load eagerly, ideally with a fetchpriority="high" hint, so the browser fetches it immediately.
Lazy loading is the flip side of that coin. Adding loading="lazy" to below-the-fold images defers their download until the user scrolls near them, which frees bandwidth for the content that matters first. The critical rule, and the one people get wrong, is to never lazy load your LCP image. Lazy loading the hero delays the very element you are being measured on and can push LCP in the wrong direction.
Finally, reserve space. Always set explicit width and height attributes (or an equivalent CSS aspect ratio) so the browser can lay out the page before the image arrives. Omitting them causes Cumulative Layout Shift, the third Core Web Vitals metric, as content jumps around during load. In my experience auditing sites, missing dimensions and lazy-loaded hero images are two of the most common and easily fixed causes of a failing CWV assessment.
On-Page Signals: Alt Text, File Names and Context
Once the file itself is optimised, the markup around it tells search engines what the image is. This is the classic image SEO layer, and it still matters, especially for ranking in Google Images.
Alt text is the most important signal. It is the primary way crawlers understand an image, and it is a real accessibility requirement for screen reader users. Write it as if you were describing the image to someone who cannot see it: be specific and accurate, include your target keyword only where it genuinely fits, and never keyword-stuff. “Woman kayaking on Lake Bled at sunrise” beats “kayak kayaking kayak tours Slovenia cheap.” For purely decorative images that carry no meaning, use an empty alt="" so assistive technology skips them.
Descriptive file names help too. red-leather-ankle-boots.webp communicates far more than IMG_4823.webp, both to Google and to anyone reading the URL. Rename files before upload rather than after, because changing them later breaks references.
Surrounding context is the underrated part. Google uses the text near an image, the caption, the heading, and the anchor text of any link to it to understand relevance. An image dropped into a page with no supporting copy is far harder to interpret than one embedded in relevant, descriptive content. Treat the image and its surrounding text as a single unit that reinforces one topic.
Getting Indexed: Image Sitemaps and Google Images
Optimising an image is pointless if Google never sees it. Most of the time a normal crawl of your HTML is enough for discovery, but several situations quietly hide images from the index.
The usual culprits are images loaded by JavaScript, images served from a separate CDN domain, and overly aggressive lazy loading that Google’s renderer never triggers. If your images live in any of those conditions, adding image information to your XML sitemap gives Google an explicit, crawlable list of every image URL and dramatically improves the odds of indexing.
For large, image-driven sites, providing image details in the sitemap is close to mandatory. Stock libraries, ecommerce catalogues, and publishers all depend on image search, and leaving discovery to chance leaves traffic on the table. Google’s Google Images documentation sets out exactly what it wants: high-quality images, descriptive alt text and filenames, images placed near relevant text, and fast-loading pages.
Beyond markup, eligibility for richer image treatment often depends on structured data. Product, recipe, and article schema can attach your images to rich results and carousels, which is why image work and structured data belong in the same workflow. If an image matters commercially, make sure it is both crawlable and described in machine-readable markup.
The 8-Step Image SEO Checklist
This is the exact sequence I run when auditing images on a site. Work through it page by page for your most important templates and you will catch the overwhelming majority of image SEO problems.
- Audit total image weight. Open the page in Chrome DevTools, filter the network tab to images, and note the total bytes. Anything over a few hundred kilobytes on a content page deserves scrutiny.
- Fix the formats. Convert photographs to AVIF or WebP with appropriate fallbacks. Convert logos and icons to SVG. Replace animated GIFs with video.
- Right-size every image. Ensure the delivered dimensions match the display size, and implement responsive
srcsetso devices download only what they need. - Compress properly. Run images through a quality-aware compressor. For most photos, a WebP at quality 75 to 80 is visually indistinguishable from the original at a fraction of the size.
- Handle the LCP image. Identify the Largest Contentful Paint element, load it eagerly with a high fetch priority, and confirm it is never lazy loaded.
- Lazy load the rest. Add
loading="lazy"to below-the-fold images to save initial bandwidth. - Write real alt text and file names. Describe each meaningful image accurately, use descriptive filenames, and set empty alt on decorative images.
- Verify indexing. Add image information to your sitemap where needed, and confirm key images are crawlable and not blocked in robots.txt.
Run this as a repeatable process, not a one-off, and bake it into your publishing workflow so new images arrive optimised. That is far cheaper than retrofitting a site full of bloated uploads later.
Image SEO rewards discipline more than cleverness. Serve modern formats at the right size, protect your Largest Contentful Paint, describe images accurately, and make sure Google can crawl them. Do those consistently and images stop being the slowest, heaviest liability on your pages and start pulling their weight in both rankings and image search. If your Core Web Vitals are failing and you suspect images are the cause, that is exactly the kind of problem a focused technical SEO engagement is built to fix.
Frequently Asked Questions
Does image SEO actually affect rankings?
Yes, in two ways. Directly, optimised images can rank in Google Images and Discover, a real traffic source for visual niches. Indirectly and more importantly, images are usually the heaviest resource on a page and the most common Largest Contentful Paint element, so they heavily influence Core Web Vitals, a confirmed ranking factor. A slow hero image quietly drags down every page it sits on.
What is the best image format for SEO in 2026?
For photographs, WebP is the safe default and AVIF is better still where supported. Google reports lossy WebP is 25 to 34 percent smaller than comparable JPEG, and AVIF usually beats WebP again. Use SVG for logos and icons, and reserve PNG for lossless transparency. In practice, serve AVIF or WebP with a JPEG or PNG fallback via the picture element or an image CDN.
Is alt text still important for SEO?
Yes. Alt text is the primary way search engines understand what an image depicts and how Google decides which image results to show. It is also an accessibility requirement for screen reader users. Describe the image specifically, include your keyword only when it fits naturally, and give decorative images an empty alt attribute so assistive technology skips them.
Do I need an image sitemap?
Only in specific cases. Google discovers most images from the page HTML during a normal crawl. You benefit from image sitemap markup when images load via JavaScript, sit behind lazy loading Google struggles to trigger, or are served from a separate CDN domain. Large image-driven sites should always provide image information in their sitemaps.
How much can image optimisation improve page speed?
Often dramatically, because images are frequently the single largest contributor to page weight. Compressing an oversized hero, serving a modern format, and sizing it correctly can cut hundreds of kilobytes and shave a full second or more off Largest Contentful Paint. In audits I regularly find pages shipping multi-megabyte images that the browser then scales down, which is pure wasted bandwidth.
