Picture this: a workshop that makes handmade wood furniture hires a professional photographer, shoots beautiful photos of its pieces, and uploads them to the website. The photos really are striking — but each one weighs 3-4 MB, the file names are machine-generated strings like DSC_0412.jpg, and not one of them has alt text. The result is predictable: product pages take six or seven seconds to load on mobile, and when someone searches Google Images for 'walnut dining table', competitors' photos show up while the workshop's own products are nowhere to be found. On top of that, the slow load time drags down its rankings in regular search results too. The owner invested in photography but never built the technical foundation needed for that investment to pay off in search.
This scene is familiar to a large share of small-business websites. Images are usually the single heaviest resource on a page, yet they are almost always the last thing anyone thinks about for SEO. In reality, image SEO — optimizing pictures for both search engines and users — earns you traffic from two separate channels at once: visual searches through Google Images and Google Lens, and better classic rankings thanks to a faster page. In this guide we'll cover image optimization in five steps: measuring where you stand today, cleaning up file names and alt text, choosing the right format and size, improving loading behavior, and making sure your images can actually be discovered.
Why is image SEO a neglected traffic channel?
Image search has long been filed under 'nice to have', but it now drives serious volume. In categories where visual intent dominates — furniture, fashion, home decor, food, jewelry, landscaping — people search with their eyes first, and the Images tab is often their first stop. Google's own figures put Lens usage at billions of searches a month, and camera-based queries like 'where can I buy this' keep growing every year. As of 2026, AI Overviews also pull images into their summaries, and the images they choose tend to come from well-optimized pages with clear context. In other words, your images aren't just an entry point in the image tab; they're a gateway into classic results and AI-powered experiences alike.
The other side of the coin is performance. HTTP Archive has shown for years that images make up the single biggest chunk of an average page's total weight. Google's page-experience metric LCP (Largest Contentful Paint) — the time it takes for the main content to render — is determined on most pages by an image loading. Hitting your Core Web Vitals targets without optimizing images is close to impossible. That's what makes image optimization such a rare kind of work: it opens a new traffic channel and strengthens your existing rankings at the same time, so a single effort pays off on two fronts.
Step 1: Take a snapshot of where you stand
The process starts with knowing what to fix. Line up three sources of data and your site's image report card writes itself:
- Search Console image performance: In the Performance report, filter the search type to 'Image'. You'll see which queries you show up for, which pages get image clicks, and your click-through rates. On most small-business sites this report is nearly empty — that's not a failure, it's proof of untapped potential.
- PageSpeed Insights warnings: Test your homepage and your two or three most important templates (product page, category page, blog post). Warnings like 'Properly size images', 'Serve images in next-gen formats', and 'Defer offscreen images' put your image-driven slowdown in front of you in MB and seconds.
- An image inventory: List every file over 1 MB in your CMS media library. Flag oversized images baked into templates (hero banners, sliders, background images) separately — since they load on every page, their impact multiplies.
Let's put a number on it. Say your category page has 12 product photos, each uploaded straight off the camera at roughly 3.5 MB. The images alone add up to 42 MB; on an average mobile connection, that page can take tens of seconds to fully load. Resize the same photos to 1200 pixels wide and convert them to WebP, and each one drops to 100-150 KB — the page's image weight falls from 42 MB to around 1.5 MB. That alone, without changing anything else, is a noticeable speed jump. For a broader audit, our technical SEO checklist covers a full crawl plan that includes images.
Step 2: Tell Google what the image is with file names and alt text
Search engines 'see' images better every year, but ranking decisions still lean heavily on text signals: file name, alt text, the content surrounding the image, and the page's title. Google Search Central's own documentation recommends using all of these signals.
Turn the file name from a camera code into a description
The file name is the first clue you give Google about what an image shows. IMG_2847.jpg says nothing; walnut-dining-table-6-seater.jpg tells the whole story at a glance. The practical rules are simple: separate words with hyphens, keep it short and descriptive, and stick to plain ASCII characters — accented letters and spaces can cause encoding problems in URLs. Renaming hundreds of existing images retroactively is rarely worth the effort; making the rule standard for every image you upload from today onward is enough.
Write alt text for people, not for Google
Alt text does two jobs: it describes the image to visitors using screen readers, and it tells search engines what the image is about. Good alt text reads like you're describing the picture to someone over the phone. Compare the two: alt="table, wood table, dining table, walnut table price" is keyword stuffing and exactly the kind of spam signal Google warns against; alt="A 6-seat walnut dining table with a live-edge top and black metal legs" is correct for both accessibility and SEO. For purely decorative images (divider lines, ornamental patterns), leaving alt text empty is also a legitimate choice — not every image needs a keyword forced into it.
Alt text doesn't work alone; Google also reads the text immediately around the image, any caption, and the page's title as context. A product photo sitting next to a paragraph that genuinely describes the product is a far stronger candidate than the same photo buried in a context-free gallery. That's why placing images near the text they relate to outperforms scattering them around for pure decoration. We cover how alt text, headings, and surrounding copy fit together in more depth in our on-page SEO guide.
Step 3: The right format and size — WebP, AVIF, and responsive images
Format matters more than most people expect. By Google's own measurements, WebP produces files roughly 25-34% smaller than JPEG at the same visual quality, and AVIF is often smaller still. As of 2026, every modern browser supports both formats, so the old 'what about older browsers' worry is largely behind us. Common platforms like WordPress can auto-convert uploaded images to WebP; on custom-built sites, an image-processing library or a CDN layer can do the same conversion on the fly.
Just as important as format is serving the image at the size it will actually display. Uploading a 4000-pixel-wide product photo that will only ever show at 400 pixels means forcing the user to download ten times more data than they need. The fix is to offer several sizes of the same image through HTML's srcset and sizes attributes and let the browser pick the one that fits the device. Modern CMSs and frameworks generate this automatically — your job is to verify the mechanism is actually working.
Make these three rules standard practice:
- Set a maximum width: Find the largest width an image can ever display at on your site (often 1200-1600 pixels for most designs) and resize originals down to that. Print-quality raw files have no business sitting on a web server.
- Lock in a quality setting: In WebP or AVIF, 75-85% quality cuts file size dramatically for product photos with no visible loss. 100% quality is wasted bandwidth on the web.
- Specify width and height in HTML: width and height attributes must always be set so the browser can reserve the image's space before it loads. This is the simplest fix for the layout jumping around while a page loads (the CLS metric).
Automate the conversion instead of doing it by hand
Converting a handful of images one by one with a free tool like Squoosh is fine to get started, but on a site that regularly publishes new content, that quickly becomes unsustainable. The lasting fix is to automate the conversion: on WordPress, image-optimization plugins compress and generate WebP versions of every upload automatically; on custom-built sites, an image-processing library like sharp or a CDN's on-the-fly conversion layer does the same job at build time or at request time. Whichever method you pick, the standard is the same: the raw photo your content team uploads should never reach the live site in its original form.
Step 4: Make loading behavior smarter — lazy loading and caching
A category page might show 40 product images, but a visitor's screen only ever displays 6-8 of them at a time. Loading the rest immediately on page load wastes bandwidth. The browser-native loading="lazy" attribute lets offscreen images load only as the user scrolls to them; it needs no extra library and is officially recommended by Google.
There's one critical exception: don't apply lazy loading to the main image visible in the first screen — often the hero or product image that determines your LCP measurement. That image should load first, not last; use fetchpriority="high" to tell the browser 'fetch this one before anything else'. A common mistake is a theme or plugin adding lazy loading to every image automatically, delaying the main one; PageSpeed Insights flags this as 'LCP image was lazily loaded'.
The second half of loading behavior is caching: images rarely change, so they should be served with long-lived Cache-Control headers and shouldn't need to be re-downloaded on repeat visits. On sites with growing traffic, serving images through a CDN cuts both load time and server load. To see the effect of these settings on LCP and overall speed metrics, record your PageSpeed scores before and after — the improvement is usually visible on the very first test.
Step 5: Guarantee discoverability — image sitemaps and structured data
Optimizing an image isn't enough; Google needs to be able to find and index it. There are four checkpoints here:
- Image sitemap: Adding image tags to your existing XML sitemap tells Google directly about the important images on each page. Google Search Central specifically recommends this for discovering images loaded later via JavaScript. A separate dedicated image sitemap file does the same job.
- Serve images as real img elements: Images shown through a CSS background-image don't get indexed in Google Images. Any image you want to rank needs to be an actual img element in the HTML.
- Add context with structured data: The image field in Product, Recipe, and Article schema helps your images show up in rich results and shopping experiences. Results that pair a product photo with price and availability information consistently get more clicks than plain image results.
- Keep image URLs stable: Every time an image's address changes, Google treats it as a brand-new image and starts evaluating it from scratch. During a CDN switch or a site migration, image URLs need to be preserved or properly redirected.
For stores with thousands of product images, these steps need to be solved at the template level, systematically, rather than one image at a time — we cover this in full in our e-commerce SEO guide.
Recap: the five-step image SEO process and realistic expectations
Let's pull the process together: first you measured where you stand with Search Console and PageSpeed; then you turned file names and alt text into real descriptions of the image; you started serving images in WebP/AVIF at the right sizes; you made loading behavior smarter with lazy loading and caching; and finally you locked in discoverability with an image sitemap and structured data. The speed side of these five steps shows up immediately — LCP improves the moment file sizes drop. Traffic from image search is a more patient game: it can take Google weeks to recrawl and reindex your images, and you'll typically see a meaningful bump in Search Console's image filter within two to three months.
On priorities: if your resources are limited, fix the speed side first — format, size, and loading behavior. That work affects your entire site and clears the way for every other SEO investment you make. Roll out alt text and discoverability work gradually, starting with the pages that drive the most traffic and revenue; trying to perfect every image on a thousand-page site all at once usually means finishing none of them.
One closing note: image SEO isn't a one-time cleanup — it's a standard that needs to become part of how you produce content. Every new image should go through naming, alt text, format, and size rules, and lazy loading and LCP settings need to be re-tested whenever you change templates. Otherwise the site you cleaned up today comes back with the same problems a year from now.
At Welda, our SEO team treats image optimization not as a standalone project but as part of the technical audit layer of our SEO consulting service: we inventory your site's images, prioritize the speed and indexing issues, and leave your team with a clear roadmap to act on. If your images are slowing your site down instead of bringing in traffic, get in touch and we'll review your current setup and build you a priority list of your own.