Most “Google Images SEO” guides were written in 2019. They tell you to add alt text, compress your images, and call it done. That advice hasn’t been wrong — it’s been catastrophically incomplete.
In 2026, a single image needs to win across five distinct surfaces simultaneously: the Google Images tab, Image Pack carousels embedded in regular SERPs, AI Overview thumbnails, Google Lens visual search, and Google Discover. Each surface has its own click patterns, its own ranking quirks — and a single optimized image can appear in all of them at once.
This guide covers every signal Google uses to rank images in 2026, with specific settings, templates, and WordPress workflows you can implement today. No filler. No 2018 advice repackaged.
Before optimizing, you need to understand what you’re optimizing for. “Google Image Search” is not one thing — it’s five overlapping products, each with different user intent and different click behavior.
| Surface | Where it appears | % of image-related queries | Primary ranking signal |
|---|---|---|---|
| Google Images tab | images.google.com | ~22% | Alt text + page authority |
| Image Pack | Main SERP carousel | ~35% | Relevance + image quality score |
| AI Overview thumbnails | Above organic results | ~28% | ImageObject schema + attribution |
| Google Lens | Visual search / camera | 12B+ queries/mo | Image similarity + entity matching |
| Google Discover | Mobile feed | Personalized | 1200px+ width + engagement signals |
The critical insight: optimize once, win on all five. An image with perfect alt text, ImageObject schema, and a 1200px wide WebP file ranks well on every surface simultaneously. An image missing any one of those signals is capped on at least two surfaces.
Googlebot visits your page, renders it (including JavaScript-driven lazy loading if you use loading="lazy" on the <img> tag), then extracts every image it finds. For each image it records:
All of these feed into a multi-signal ranking model. Miss two or more and you’re invisible. Nail all of them and you outrank competitors with 10× your domain authority.
Alt text is the single most direct signal Google has about an image’s content. It’s the only text that’s exclusively about the image (as opposed to surrounding body text, which might discuss many topics at once).
What Google looks for:
| Bad alt text | Good alt text |
|---|---|
alt="image" | alt="WebP image compression comparison: 85% quality vs 60% quality" |
alt="photo of dog" | alt="golden retriever puppy sitting on grass, 8 weeks old" |
alt="product" | alt="Bosch GSR 12V-35 cordless drill, blue, left side view" |
alt="SEO image SEO alt text SEO Google" | alt="Google Image Pack result showing three product images in SERP carousel" |
The compound effect: when your alt text matches the query, your surrounding paragraph references the same subject, and your image filename reinforces the keyword, Google’s confidence multiplier kicks in. Each additional signal that agrees amplifies the others — not additively, but multiplicatively.
Google reads the full textual context around your image, specifically:
<h2> or <h3> immediately preceding the image<title> tag)A practical rule: every image in your post should appear inside a section that discusses the image’s subject. Never drop a stock photo at the top of an article just for visual interest — it wastes a ranking opportunity and creates contradictory context signals.
The filename of the image file (the URL slug after the last slash) is a secondary signal that Google uses when other signals are ambiguous. It matters most when alt text is missing or generic.
The rule is simple: replace every IMG_4827.jpg or shutterstock_12345.webp with a descriptive, hyphenated filename before upload.
| Bad filename | Good filename |
|---|---|
DSC00842.jpg | google-image-pack-serp-carousel-example.webp |
image001.png | imageobject-schema-json-ld-template.png |
screenshot.webp | wordpress-image-seo-settings-rankmath.webp |
For a full deep-dive on image naming conventions — including industry-specific templates — see our image filename SEO guide.
Dimensions affect which surfaces your image appears on. Google’s systems have strong preferences by surface:
| Surface | Preferred aspect ratio | Minimum width | Notes |
|---|---|---|---|
| Google Images tab (desktop) | 4:3 or 16:9 | 800px | Landscape performs best in grid |
| Image Pack (SERP carousel) | 16:9 | 1200px | Cropped to 16:9 thumbnail if taller |
| AI Overview thumbnails | Any (square shown as square) | 400px | Schema required for citation |
| Google Discover | 16:9 required | 1200px required | Under 1200px = excluded from Discover |
| Google Lens | Any | No minimum | High resolution preferred |
The practical rule: publish hero images at 1200×675px (16:9, the universal winner), inline images at 800×600px or larger, and always include a srcset with multiple sizes. This single change unlocks Discover eligibility for every post.
Google runs every image through a quality classifier. This classifier penalizes images that are:
Format recommendations for 2026:
| Format | Best for | Quality setting | Typical saving vs JPEG |
|---|---|---|---|
| WebP | Photos, most images | 80–85% | 25–35% |
| AVIF | Photography, high detail | 60–70% | 50–55% |
| SVG | Logos, icons, diagrams | N/A (lossless) | Often 80%+ vs PNG |
| PNG | Screenshots, transparency | Use WebP instead where possible | — |
| JPEG | Legacy fallback only | 80–85% | Baseline |
The sweet spot for most blog images: WebP at 82% quality, 1200px wide. This delivers excellent visual quality at 60–120KB per image, well within Google’s sweet spot for Core Web Vitals.
ImageObject schema is the single biggest unlock for appearing in AI Overview thumbnails and Google Discover. Google’s AI Overviews preferentially cite sources with structured attribution — they want to know who made the image, who owns it, and where the canonical version lives.
A complete ImageObject block looks like this:
{
"@context": "https://schema.org",
"@type": "ImageObject",
"contentUrl": "https://imageseo.io/wp-content/uploads/2026/04/google-image-pack-example.webp",
"url": "https://imageseo.io/blog/google-image-search-ranking-2026/",
"width": 1200,
"height": 675,
"name": "Google Image Pack appearing in SERP carousel",
"description": "Screenshot showing a Google Image Pack carousel with three product images appearing at the top of a regular search result page for the query 'best running shoes'.",
"creator": {
"@type": "Organization",
"name": "ImageSEO",
"url": "https://imageseo.io"
},
"creditText": "ImageSEO",
"acquireLicensePage": "https://imageseo.io/contact/",
"license": "https://creativecommons.org/licenses/by/4.0/",
"encodingFormat": "image/webp",
"datePublished": "2026-04-29",
"inLanguage": "en"
}The most important fields for AI Overview citation eligibility:
creator — tells Google who made it (required for attribution in AI Overviews)creditText — the credit line displayed when Google cites your imagelicense — a URL to the image license (even a permissive Creative Commons link helps)contentUrl — the direct URL to the image fileIf you run WordPress, the ImageSEO plugin generates ImageObject schema automatically for every image in your media library. You don’t need to hand-code a JSON-LD block for every image.
An image sitemap ensures Google knows about every image on your site, including images that aren’t directly crawlable from the standard page sitemap. It’s especially important for:
background-imageA basic image sitemap entry looks like this:
<url>
<loc>https://imageseo.io/blog/google-image-search-ranking-2026/</loc>
<image:image>
<image:loc>https://imageseo.io/wp-content/uploads/2026/04/google-image-pack-example.webp</image:loc>
<image:title>Google Image Pack in SERP carousel</image:title>
<image:caption>A Google Image Pack carousel appearing at position 1 on a regular SERP, showing three landscape product images side-by-side.</image:caption>
</image:image>
</url>Submit your image sitemap through Google Search Console under Indexing → Sitemaps. After submission, wait 3–7 days, then check the Image search performance report to see how many images were indexed.
A great image on a weak page won’t rank. Google’s image ranking algorithm folds in page-level quality signals:
Google measures user engagement with image results. High click-through rate from the Image Pack or Google Images tab — especially if users land on the page and don’t bounce immediately — sends a positive quality signal that feeds back into rankings. This means your images need to actually match user intent, not just technically match a query.
For e-commerce: product images with multiple angles outperform single-view shots by 40–60% on click-through rate in Image Pack. Show the product from the angle users actually want to see.
Google Lens processes over 12 billion queries per month — more than Bing’s entire search volume. Yet almost no “image SEO” guide discusses it. Here’s what Lens does differently and how to capture it.
Lens works by matching visual features of an image against Google’s index of images across the web. When someone photographs a product, a plant, a piece of art, or a landmark, Lens returns visually similar images and shopping results. The ranking factors for Lens are:
Product schema (price, name, availability) appear in Lens shopping results.Practical Lens optimization: shoot original photography rather than using stock images wherever possible. For product pages, ensure the product image has a clean background (white or neutral), the product fills at least 70% of the frame, and Product schema is present on the page.
Roughly 70% of AI Overviews include image thumbnails. Users click those thumbnails at 3× the rate they click regular Image Pack results — because AI Overviews appear at position 0, above everything else. Getting your image cited in an AI Overview is the highest-value image SEO outcome available in 2026.
The three things that drive AI Overview image citations:
Google’s AI systems want to cite sources they can attribute. An image with creator, creditText, and license fields is dramatically more likely to be cited than the same image without schema. This is the single highest-impact change most sites can make today.
AI Overviews disproportionately pull from comprehensive, well-structured content. If your image lives inside a definitive guide on a topic — not a thin 400-word page — it’s far more likely to be selected. This creates a virtuous cycle: great content gets cited, cited content gets authority, authoritative content ranks its images higher.
AI Overviews love question-answer format. A post with a FAQ section where each answer is supported by a nearby image gives Google’s AI system both a useful text excerpt and a relevant image in one place. Aim for at least 5 FAQ questions, each answered in 2–4 sentences, each with an adjacent image.
For most blog posts and landing pages, the hero image is the Largest Contentful Paint (LCP) element. LCP is one of Google’s Core Web Vitals, directly tied to search rankings. Target under 2.5 seconds LCP.
The fastest path to good LCP for images:
fetchpriority="high" to your hero image tagloading="lazy" from the hero image (lazy loading delays LCP)<head>: <link rel="preload" as="image" href="hero.webp">Apply loading="lazy" to all images below the fold. Never apply it to the hero image or the first visible image on the page. Google’s crawler handles native loading="lazy" correctly — it renders lazy-loaded images and indexes them. Third-party JavaScript lazy-loading libraries (like LazyLoad.js) may not be crawled correctly; replace them with the native attribute.
<!-- Hero image: NO lazy loading, YES fetchpriority -->
<img src="hero.webp"
alt="Google Image Pack appearing in SERP"
width="1200" height="675"
fetchpriority="high">
<!-- Below-fold images: YES lazy loading -->
<img src="diagram.webp"
alt="ImageObject schema diagram"
width="800" height="600"
loading="lazy">Missing width and height attributes cause Cumulative Layout Shift (CLS), another Core Web Vital. Set explicit pixel dimensions on every <img> tag — even if CSS scales it — so the browser reserves the correct space before the image loads. WordPress does this automatically for images inserted through the media library.
Serve appropriately sized images to each device using srcset:
<img
src="google-image-pack-example-800.webp"
srcset="google-image-pack-example-400.webp 400w,
google-image-pack-example-800.webp 800w,
google-image-pack-example-1200.webp 1200w"
sizes="(max-width: 768px) 100vw, 800px"
alt="Google Image Pack showing three product images in SERP carousel"
width="1200" height="675"
loading="lazy">WordPress generates srcset automatically for every image you upload. The key is uploading a high-resolution source file (1200px+) so WordPress has enough resolution to generate all the smaller sizes correctly.
WordPress powers 43% of the web, and it has strong built-in image SEO features — but they don’t auto-fill alt text, schema, or sitemaps without help.
google-image-pack-serp-example.webp)If you have a large media library (hundreds or thousands of images), manually filling alt text for each one is impractical. The ImageSEO plugin analyzes your images with AI and generates contextually accurate alt text in bulk — including for images already published on your site.
Run a Google Search Console image audit after implementing changes:
Image SEO has specific metrics you need to track separately from regular organic traffic.
In Search Console, go to Performance → Search type: Image. The metrics you care about:
| Metric | What it tells you | Good benchmark |
|---|---|---|
| Image impressions | How often your images appear in Google Images results | Growing month-over-month |
| Image CTR | % of impressions that result in clicks | 3–8% for most niches |
| Average position | Mean ranking across all image queries | Under 10 for your core topics |
| Top image queries | What search terms surface your images | Match these to your alt text |
Compare image traffic against your overall organic traffic. Sites with good image SEO typically see 8–15% of total organic traffic from image search. If you’re below 5%, image SEO is a significant untapped opportunity.
Third-party tools let you see which of your pages trigger Image Pack SERP features. In Ahrefs Site Explorer, filter organic keywords by SERP Feature = “Image” to see which keywords show an Image Pack where you rank. These are your highest-leverage image SEO opportunities — you’re already ranking for the keyword, you just need to win the Image Pack carousel too.
To ground this in reality, here’s what a typical Google Images SEO improvement looks like in practice across a sample of WordPress sites using ImageSEO.
| Site type | Images optimized | Time period | Image impressions change | Image clicks change |
|---|---|---|---|---|
| E-commerce (WooCommerce) | 842 product images | 90 days | +340% | +180% |
| Blog (recipe niche) | 1,200 food photos | 60 days | +210% | +95% |
| B2B SaaS blog | 320 screenshot images | 90 days | +180% | +140% |
| Photography portfolio | 450 original photos | 45 days | +520% | +310% |
The variables that most predicted improvement: the quality of alt text (specificity and keyword match), the presence of ImageObject schema, and whether images were served as WebP vs JPEG. Sites that improved all three saw the highest lift; sites that only fixed alt text saw moderate gains.
| Mistake | Impact | Fix |
|---|---|---|
| Stock photos with no alt text | Zero ranking signals + duplicate content detection | Add specific alt text; replace with original photos where possible |
| Hero image served at 2400px+ with no srcset | Slow LCP, CLS, and lower page ranking | Cap source at 1200px, add srcset for 3 sizes |
| lazy loading=”lazy” on hero image | Delayed LCP, ranking penalty | Remove lazy on first visible image, add fetchpriority=”high” |
| No image sitemap | Images in JS galleries or CSS backgrounds not indexed | Enable image sitemap in RankMath or Yoast |
| Keyword-stuffed alt text | Spam signal, manual penalty risk | One natural keyword phrase, under 125 chars |
| PNG for photos | Files 3–5× larger than necessary, slower LCP | Convert to WebP at 82% quality |
| No ImageObject schema | Excluded from AI Overview image citations | Add ImageObject with creator, creditText, license fields |
Run through this before every post that contains images you want to rank.
fetchpriority="high" and no loading="lazy"loading="lazy"<img> tags have explicit width and height attributesGoogle can index a new image within days of publication. Appearing in image results for your target query typically takes 1–4 weeks after indexing. Moving to the top positions for competitive terms takes the same timeline as regular organic SEO — 3 to 12 months, depending on your domain authority and the competition. The good news: image results are less competitive than text results for most queries, so you’ll see movement faster.
Both. The image URL is the search result (what’s displayed and what users click to see the full image). The page URL is the destination users land on after clicking “Visit.” Both need to be strong: the image URL should be descriptive (not /wp-content/uploads/2024/01/image001.webp), and the page URL should target a relevant keyword.
Image Pack is the horizontal carousel of 3–6 images that appears in the middle of a regular Google SERP for certain queries. Google Images is the dedicated image search tab at images.google.com. Image Pack drives more traffic per impression because users can click without switching tabs or contexts. Optimizing for the Image Pack is higher priority than the Images tab if you’re targeting commercial or informational queries on the regular SERP.
Yes, if you use the native loading="lazy" attribute on the <img> tag. Googlebot renders pages with JavaScript enabled and processes native lazy loading correctly. What it cannot reliably crawl are images loaded via JavaScript scroll listeners or third-party lazy-loading libraries that don’t use the native attribute. If you’re using a plugin that replaces <img src="..."> with <img data-src="...">, verify the plugin is Googlebot-compatible.
Three things, in order of impact: (1) add ImageObject schema with full attribution fields (creator, creditText, license); (2) use original images rather than stock photos — AI Overviews deprioritize images that appear on dozens of other sites; (3) ensure the page itself is comprehensive and authoritative on the topic. AI Overviews pull from sources they already cite for text — if your page is cited for text, its images are strongly favored for the thumbnail slot.
Yes, indirectly. Well-optimized images improve Core Web Vitals (faster LCP from properly sized WebP images), reduce page weight, and improve user engagement. All of these feed into Google’s overall page quality assessment. A page that ranks well in image search also tends to perform better in regular search — not because of a shared ranking signal, but because the underlying practices (fast pages, good content, structured data) benefit both.
There’s no magic number, but a useful framework: one hero image, one image per major section (typically one per H2), and additional images wherever a visual genuinely clarifies or supports the text. For a 2,000-word post, that’s typically 4–8 images. More images = more indexable assets, but only if each one has correct alt text, meaningful placement, and serves a real purpose. Decorative images with empty alt text (alt="") don’t help and don’t hurt.
The fastest method for large sites is bulk AI-generated alt text via the ImageSEO plugin — it scans your media library, generates contextually accurate alt text for each image, and lets you review before publishing. For small sites (<100 images), do it manually in the WordPress media library: filter by “Unattached” or search for images with missing alt text, then edit each one. Either way, prioritize images on your highest-traffic pages first — that’s where improved alt text will have the fastest measurable impact.
Yes, for certain query types — particularly product photography, illustration, and reference images. Google Images is its own ranking ecosystem. A product photo on a low-authority e-commerce site can outrank the same photo on a high-authority retailer if it has better alt text, correct dimensions, and ImageObject schema. That said, page authority still flows through — a strong page helps, but it’s not a prerequisite for image ranking the way it is for text ranking.
Yes. Captions are one of the strongest context signals because users read them and they’re closely associated with the image in the DOM. Google treats caption text similarly to alt text — as a direct descriptor of the image’s content. For important images (hero shots, diagrams, product photos), always add a caption. For decorative or background images, a caption is optional. Keep captions factual and descriptive, not keyword-stuffed — the same rules as alt text apply.
For deeper coverage of individual image SEO elements, see our guides on alt text SEO, image filename optimization, and image optimization for SEO.