Most publishers say they care about traffic, but then they treat images like decoration. That is sloppy thinking. Google’s own documentation is clear that the same image best practices help images appear across Google Search, Google Images, and Google Discover. In other words, images are not just design assets. They are part of search visibility, click appeal, and content understanding. If your images are weak, irrelevant, badly labeled, or too small, you are cutting your own reach before the article even gets judged on quality.
This matters more in Discover because Discover is a browsing surface, not a search-results page where users are already motivated by a typed query. Google explicitly recommends compelling, high-quality, relevant large images for Discover and says large images are more likely to generate visits. That means your headline and your image work together as the first sales pitch. If the visual is generic, text-heavy, poorly cropped, or tiny, you are not “doing okay.” You are wasting impressions.
What Google Actually Wants From Your Images
Google’s image SEO guidance is not mysterious. It wants images that it can discover, index, understand, and confidently show to users. That starts with using standard HTML image elements. Google says it can find images in the src attribute of an <img> element, but it does not index CSS background images the same way. A surprising number of publishers still rely on design-heavy implementations that look good to humans but reduce clarity for crawlers. That is not advanced SEO. That is unforced damage.
Google also wants context. Its documentation says Google extracts information about an image from the page content, including captions and image titles, and recommends placing images near relevant text on pages that are relevant to the subject of the image. It also says filenames should be short but descriptive, and that alt text is one of the most important attributes for supplying image metadata. Alt text helps accessibility, but Google also uses alt text together with computer vision and page content to understand the image. So anyone still stuffing keywords into alt text or uploading files named IMG00023.jpg is doing lazy work, not SEO.
The Discover Image Rules Publishers Keep Missing
Discover is less forgiving because visual presentation matters more there. Google’s Discover documentation recommends images that are at least 1200 pixels wide, have more than 300,000 total pixels, and ideally fit a 16:9 ratio because Google may crop them for Discover. It also says large image previews must be enabled by max-image-preview:large or by using AMP. On top of that, Google says you should use either schema.org markup or the og:image meta tag to specify a large, relevant, representative image, because that can influence which image is chosen as the thumbnail in Discover.
That last point matters because many publishers still assume that uploading any featured image is enough. It is not. Google specifically says to avoid generic images like logos in schema or og:image, and to avoid text-heavy images there as well. It also advises against extreme aspect ratios and recommends high resolution when choosing preferred images. So if your Discover thumbnail is bad, the problem may not be “Google chose the wrong image.” The real problem may be that you gave Google weak options and poor metadata.
Table: What Good Image SEO Looks Like Versus What Publishers Commonly Do
| Image SEO area | What Google recommends | Common publisher mistake | Why it hurts |
|---|---|---|---|
| Image size for Discover | At least 1200 px wide and over 300,000 total pixels | Small featured images or compressed thumbnails | Reduces chance of strong large-image presentation in Discover |
| Image markup | Use <img src> and provide fallback URL for responsive images |
CSS background images or broken responsive handling | Makes discovery and indexing less reliable |
| Preferred thumbnail signal | Use schema.org markup or og:image with a relevant image |
Letting social image defaults or logos fill the slot | Weakens thumbnail selection and click appeal |
| Alt text | Useful, specific, context-aware description | Keyword stuffing or missing alt text | Hurts accessibility and image understanding |
| Filenames | Short, descriptive names | image1.jpg, final-final.jpg, random camera names |
Gives Google weaker clues about the image subject |
| Page context | Put images near relevant text and captions | Decorative images unrelated to the section | Makes the image less semantically useful |
| Metadata for rights | Add structured data or IPTC photo metadata if relevant | No creator or licensing information | Misses eligibility for enhancements like the Licensable badge |
Why Image Metadata Matters More Than Most Publishers Realize
When people hear “metadata,” they often think of some technical extra that only large media brands care about. That is wrong. Google’s image metadata documentation says structured data or IPTC photo metadata can help Google show more details about an image, including creator, credit, and licensing information. It also says licensing information can make an image eligible for the Licensable badge in Google Images. That may not directly guarantee Discover traffic, but it improves how Google understands and presents visual assets, especially for publishers, photographers, and media-heavy websites.
Google also explains that structured data should be added for each image instance on each page where that image appears, while IPTC metadata is embedded in the image itself and stays with the file. It further notes that if structured data and IPTC metadata conflict, Google will use the structured data information. That means publishers need consistency. Slapping random image details into plugins without checking output is how sites end up with messy signals and unreliable presentation.
The Simple Fixes That Usually Make the Biggest Difference
The first fix is to stop choosing featured images based only on visual taste. Pick images that are representative of the article, wide enough for Discover, high enough in resolution, and usable in landscape cropping. Google explicitly recommends large images, warns against extreme aspect ratios, and notes that Discover may crop images automatically. So your job is not just to upload a pretty picture. Your job is to upload one that survives cropping and still communicates the article instantly.
The second fix is to clean up image context. Google says it uses captions, image titles, page content, and alt text to understand images. So the image should sit near the most relevant text, the alt text should describe the image naturally, and the filename should make basic sense. None of this is glamorous, but most SEO gains come from fixing dumb habits, not chasing magical hacks. A descriptive image next to relevant copy will usually outperform a random stock photo with a stuffed alt tag.
The third fix is technical consistency. Use crawlable image URLs, standard HTML image elements, supported formats, and image sitemaps where helpful. Google supports formats such as JPEG, PNG, WebP, SVG, and AVIF, and recommends submitting image sitemaps to help it discover image URLs it might not otherwise find. It also recommends verifying CDN ownership in Search Console if you host images on a CDN so crawl issues can be surfaced properly. If your image delivery setup is fragmented, you are making your own life harder.
What Publishers Should Stop Ignoring in 2026
The brutal truth is that image SEO is now tied to visibility, user experience, and click behavior more directly than many publishers want to admit. Google’s documentation for Discover says to use compelling, high-quality images, avoid misleading preview content, and provide a great page experience. It also says Discover traffic is inherently less predictable than keyword-driven search traffic. That means you cannot control Discover fully, but you can absolutely improve your chances by fixing image quality, size, relevance, and metadata. Pretending images are secondary is just lazy publishing strategy.
Publishers also need to stop confusing image SEO with only alt text. Alt text matters, yes, but Google’s guidance is broader: crawlable images, proper HTML embedding, responsive handling, strong on-page context, descriptive filenames, structured data where relevant, and thumbnail-friendly image choices all matter. Good image SEO is a system, not a checkbox. If one part is weak, the result usually looks weak too.
A Practical Image SEO Workflow for Better Discover and Search Performance
A smart workflow starts before upload. Decide what image best represents the page, crop it in a way that still works in landscape, make sure it is at least 1200 pixels wide, and avoid loading the image with unnecessary text. Then give the file a clean descriptive name and pair it with useful alt text. After that, make sure the image is inserted with proper HTML image markup and is located near the most relevant text on the page.
After publishing, confirm that your page is accessible to Google, your images are crawlable, your sitemap is updated, and your preferred image is declared in schema or og:image when appropriate. If licensing or creator details matter to your content model, add structured data or IPTC metadata as well. This is not over-optimization. It is basic competence for any publisher who claims to care about traffic from visual surfaces.
Conclusion
Image SEO for Discover and Google is not optional polish anymore. It affects how Google discovers your visuals, understands them, selects thumbnails, and presents your content across Search, Images, and Discover. Google’s own documentation makes the priorities obvious: use crawlable image markup, choose relevant high-resolution images, enable large previews for Discover, write useful alt text, use descriptive filenames, and avoid generic or text-heavy thumbnail candidates.
The bigger issue is not lack of guidance. It is that many publishers still ignore basic image discipline and then blame traffic volatility. That is weak thinking. If you want better visibility, stop treating images as decoration and start treating them like ranking and click assets. That is where the real improvement begins.
FAQs
Does image SEO really help with Google Discover?
Yes. Google explicitly says the same general image recommendations help images appear in products such as Google Search, Google Images, and Google Discover, and it specifically recommends compelling, high-quality, large images for Discover.
What image size does Google recommend for Discover?
Google recommends images that are at least 1200 pixels wide, have more than 300,000 total pixels, and work well in a 16:9 format for Discover. Large previews also need max-image-preview:large or AMP.
Is alt text enough for image SEO?
No. Alt text is important, but Google also looks at page context, captions, image titles, filenames, structured data, crawlability, markup method, and thumbnail signals such as schema or og:image.
Can Google index CSS background images?
Google’s documentation says it can find images in the src attribute of <img> elements, but it does not index CSS images. That is why standard HTML image embedding matters.
What is image metadata in Google’s system?
Google says image metadata can include creator, credit, and licensing details, and that you can provide it through structured data or IPTC photo metadata. In some cases, this can make images eligible for enhancements such as the Licensable badge.
Should publishers use logos as featured images for Discover?
No. Google specifically advises against using generic images such as a site logo in schema markup or og:image, and also recommends avoiding text-heavy images there.
Click here to know more.