How does Google rank images?
Google ranks images primarily by page context — the surrounding text, captions, headings and page topic — combined with image-specific signals: descriptive filenames, alt text, and relevance. The image itself is analyzed by computer vision, but context dominates: a perfectly-named image on an irrelevant page won't rank, while a decent image on the definitive page about its subject will.
The ranking inputs, roughly ordered: page relevance (is the hosting page the best result for the query?), contextual text (captions, nearby paragraphs, headings describing the image), filename and alt text (explicit labels), image quality signals (original vs stock, resolution appropriateness), and page experience (fast, mobile-friendly pages rank their images better too).
This means image SEO is mostly *page* SEO with image-specific finishing: the image ranks where the page deserves to rank. A product photo on the web's best guide to that product outranks the same photo on a thin affiliate page — Google Images is a window into page quality, not a separate lottery.
Computer vision does matter at the margins: Google identifies objects, text (OCR) and even image uniqueness. Original images outperform stock photos — not because of a 'stock penalty', but because originals appear on pages with original content, while stock appears on thin pages. Correlation, but a useful one: original photography is an image-SEO advantage.
Traffic reality check: Google Images drives real visits for visual queries (products, diagrams, photography, memes, infographics) and near-zero for others. Prioritize image SEO where your audience *searches visually* — ecommerce, tutorials, travel, food, design — and treat it as hygiene elsewhere.
How do you name files and write alt text?
Name files descriptively in lowercase with hyphens (blue-cotton-kurta-front.jpg, not IMG_4829.jpg) and write alt text describing the image for someone who can't see it — concise, specific, no 'image of' prefix, no keyword stuffing. Filenames help Google; alt text helps blind users and Google; both help when images fail to load.
Filenames: descriptive, lowercase, hyphen-separated, in the page's language. 'mens-leather-wallet-brown-rfid.jpg' tells Google everything; 'IMG_4829.jpg' tells it nothing. Rename *before* uploading — renaming after means updating references everywhere.
Alt text serves three masters: screen readers (accessibility — the legal and moral priority), Google (relevance signal), and broken-image fallback (users see the text when the image fails). Write it as a concise description: alt='Brown leather bifold wallet with RFID blocking, open showing card slots'. Rules:
- Describe, don't keyword-stuff. 'Brown leather wallet' — not 'wallet, leather wallet, best wallet, buy wallet'.
- No 'image of' / 'picture of'. Screen readers already announce it's an image.
- Be specific. 'Golden retriever puppy sleeping on a blue rug' beats 'dog'.
- Empty alt (alt='') for decorative images. Pure decoration should be silent to screen readers — empty alt is correct, missing alt is an error.
- Keep it under ~125 characters. Screen readers may cut longer text.
Captions and surrounding text amplify everything: a caption under the image ('Figure 3: LCP improvement after hero optimization') is weighted heavily and helps readers too. Place images near the text they illustrate — Google associates nearby content with the image.
Audit existing media libraries: WordPress stores alt text per attachment — bulk-review your top pages' images and fix the worst offenders (empty alts on informative images, 'IMG_' filenames on ranking pages). The image resizer won't fix naming, but it's handy when preparing properly-sized replacements.
How do you optimize images for speed without losing quality?
Compress every image to 80–85% quality (60–90% smaller, visually identical), convert photos to WebP, serve responsive sizes via srcset, lazy-load below-the-fold images, and set explicit dimensions to prevent layout shift. Images are typically 50–80% of page weight — this is usually the largest performance win available.
The speed checklist per image:
- Compress. 80–85% JPG/WebP quality — the universal sweet spot. Use the image compressor for existing files; set upload-time compression going forward.
- Modern formats. WebP for photos (25–35% smaller than JPG); PNG only for graphics needing lossless/text sharpness. AVIF where supported for maximum savings.
- Right-size. Serve the display size — a 4000px upload in an 800px slot is pure waste. WordPress srcset handles multiples automatically if originals are sensibly sized.
- Responsive (srcset/sizes). Let the browser pick the appropriate size per device — mobile users shouldn't download desktop-sized images.
- Lazy-load below the fold. loading='lazy' (WordPress does this natively) — but never lazy-load the LCP/hero image.
- Dimensions always. width/height or aspect-ratio on every image — prevents CLS layout jumps (see the Core Web Vitals guide).
Format decision shortcut: photo → WebP/JPG; logo/icon/graphic with text → PNG (or SVG for vector logos — infinitely scalable, tiny); screenshot needing sharp text → PNG; screenshot where size matters → JPG at 80%. Convert between them with the image converter, PNG to JPG and WebP to JPG tools.
Measure the payoff: total page weight before/after (aim for <1.5 MB on content pages), LCP improvement (the hero image is usually the LCP element), and field-data vitals over the following month. Image optimization routinely delivers the biggest single performance improvement in an audit.
What structured data helps images rank?
Product markup with image arrays powers product rich results; ImageObject can specify license and creator for licensable images (earning the 'Licensable' badge in Google Images); and Article markup associates images with content. Mark up honestly — structured data describes visible content, and the Licensable badge specifically helps photographers and stock creators get credited and contacted.
Relevant schema types:
- Product + image. Product markup's image field feeds product rich results — price, availability and image together in search. Multiple angles as an image array.
- ImageObject (licensable). For creators: license URL, creator, credit — earns the 'Licensable' badge in Google Images, which filters for buyers and drives licensing inquiries. Photographers and marketplaces should implement this.
- Article/BlogPosting image. The representative image field — also feeds social sharing and Discover.
- VideoObject thumbnail. For video pages — the thumbnail is a ranking surface in video search.
Validate everything in Google's Rich Results Test — invalid markup is invisible markup. And keep the honest-markup rule absolute: structured data must describe what's visibly on the page. Fake product markup (prices that don't exist, reviews you invented) is a manual-penalty-grade violation.
Sitemaps: image sitemaps (or image extensions in your XML sitemap) help Google discover images, especially those loaded via JavaScript galleries. Most SEO plugins generate these automatically — verify yours does, particularly for portfolio and ecommerce sites where images *are* the content.
How do you handle image SEO for ecommerce and portfolios?
For stores: unique product photos (multiple angles, lifestyle, scale), descriptive filenames with product attributes, alt text per image, Product schema with image arrays, and fast loading — image search becomes a discovery channel. For portfolios: watermark judiciously (watermarks don't prevent theft, they just degrade UX), use Licensable markup, and protect with context (your site should be the canonical source).
Ecommerce image SEO:
- Original photography wins. Supplier-provided images appear on fifty competing listings; your own photos differentiate in image search and convert better on the page.
- Multiple angles, one naming system. product-sku-front.jpg, product-sku-angle.jpg, product-sku-detail.jpg, product-sku-lifestyle.jpg — systematic names scale across catalogs.
- Alt text per image, describing that specific view — not the same alt copied across all six.
- Zoom without weight. Thumbnails load fast; full-size on click — don't serve 3000px images as thumbnails.
- Image search as a channel. Track Google Images traffic in Search Console's search-type filter — for visual products it's often 10–30% of organic discovery.
Portfolio/creative sites: your images are the product. Use Licensable ImageObject markup, keep resolution web-appropriate (enough to impress, not enough to print from), and make your site the canonical source with context Google can't get from scrapers (process shots, descriptions, client stories). Watermarks: small and corner-placed if you must — giant center watermarks punish your own visitors more than thieves.
Hotlink protection (blocking other sites from embedding your images) saves bandwidth but doesn't help SEO — and image search *wants* your images found. Prioritize discoverability; use licensing metadata for protection instead.
What are the biggest image SEO mistakes?
The biggest mistakes: 5 MB uncompressed uploads, 'IMG_0042.jpg' filenames, empty or stuffed alt text, no responsive sizes (mobile downloads desktop images), lazy-loading the hero image, and decorative images missing empty alt. Each is a one-time fix per template with permanent payoff — image SEO is high-leverage precisely because everyone neglects it.
The audit hit-list — check your top pages for each:
- Weight. Any image over ~300 KB on a content page is suspect; over 1 MB is a bug. Compress and convert.
- Naming. Bulk-rename the worst 'IMG_/DSC_/screenshot' files on important pages.
- Alt. Empty alts on informative images; keyword-stuffed alts; missing alts entirely. Fix in the media library.
- Responsive. One giant file served to all devices — implement srcset.
- Hero handling. Lazy-loaded LCP image (delays your most important metric); missing preload; missing dimensions.
- Decorative noise. Spacer/divider images with filenames and alt text — give them empty alt and forget them.
- Text in images. Infographic text invisible to Google — transcribe key points as HTML text nearby; the image keeps the visual, the text gets indexed.
The compounding case: image fixes improve speed (rankings + conversions), accessibility (wider audience + legal safety), image search traffic (new discovery channel) and page experience (vitals) simultaneously. Few SEO tasks pay on four fronts at once — which is exactly why image SEO belongs early in every optimization program, not as an afterthought.
Start today: pick your ten most-visited pages, run their images through this checklist, and measure page weight and LCP before and after. The results usually sell the full-site rollout by themselves.
Compress, convert and resize images free in your browser — the first step of every image SEO fix.
Open the image compressor ↗Frequently asked questions
Does alt text directly improve rankings?
It's a relevance signal that helps images (and pages) rank for descriptive queries, and it's essential for accessibility. Write honest descriptions — alt text is for users first, and keyword-stuffed alt text helps no one.
Should I use WebP for all images?
For photos, yes — 25–35% smaller than JPG at equal quality with universal modern browser support. Keep PNG for graphics needing lossless sharpness (logos, text-heavy screenshots) and SVG for vector logos.
How many images should a page have?
As many as serve the reader — no SEO quota. But every image must earn its kilobytes: compress, size correctly, name and describe it. Ten optimized images outperform three 5 MB ones on every metric.
Can Google read text inside images?
Google OCRs image text to some degree, but don't rely on it — put important text (infographic points, diagram labels) as HTML text near the image. The image keeps the visual appeal; the text gets reliably indexed.
Do image sitemaps still matter?
They help discovery, especially for images loaded via JavaScript or on pages Google crawls infrequently. Most SEO plugins generate them automatically — verify yours includes images, particularly for ecommerce and portfolio sites.
Will compressing images hurt their ranking?
No — sensible compression (80–85% quality) is visually identical and Google doesn't penalize optimized images. It helps: faster pages rank better and users engage more. Over-compression that visibly degrades quality is the only risk, and it's easily avoided with previews.