How to Make Google Discover Use the Right Page Image

To help Google Discover use the right image, choose a relevant, high-quality landscape image for the page, identify it consistently with og:image and/or schema.org image markup, allow large previews with max-image-preview:large, and make sure Google can discover and access the image. These signals can influence Google’s automated choice; they do not force a particular thumbnail.

1. Choose an image that represents the page

Use a large image that clearly relates to the page’s main subject. A generic site logo or text-heavy graphic is usually a poor representative. Compose it for a landscape presentation: Google recommends an image at least 1200 pixels wide and more than 300,000 pixels total. A 1280 × 720 image has 921,600 pixels and meets those dimension recommendations.

Discover presents images in a 16:9 format and may crop them automatically. Keep the main subject near the center and check that it remains visible if the image is cropped. If the page has several possible images, make the intended lead image clearly representative and use that same URL in the metadata signals below.

These dimensions and composition recommendations improve eligibility and presentation; they do not guarantee that a page will appear in Discover or that Google will select a particular image. See Google’s Discover guidance.

2. Set a preferred image in page metadata

Add an Open Graph image tag to the page’s HTML head. Replace the sample URL with an absolute, publicly accessible URL for the image you chose.

<head>
  <meta name="robots" content="max-image-preview:large">
  <meta property="og:image" content="https://example.com/images/article-hero-1280x720.jpg">
</head>

You can also identify the image with schema.org structured data. For a page about an article, use the image on the page or article entity as appropriate. This minimal WebPage example shows primaryImageOfPage:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "url": "https://example.com/article",
  "primaryImageOfPage": "https://example.com/images/article-hero-1280x720.jpg"
}
</script>

Google documents both og:image and schema.org image markup as sources for image thumbnails in Search and Discover. You may use either or both. If you supply both, point them to the same representative image so the page sends a consistent signal. Google does not document one format as universally better.

Google clarified this guidance in its documentation updates on March 2, 2026. For implementation details, see Google image SEO best practices.

3. Allow a large image preview

The max-image-preview robots directive controls the maximum image preview Google may show. Add this to the page head when you want to permit a large preview:

<meta name="robots" content="max-image-preview:large">

The values are:

Value	Effect
large	Allows a larger image preview, up to the width of the viewport.
standard	Allows a default-sized preview.
none	Disallows image previews.

Check for conflicting page-level robots directives. A setting that limits previews can prevent the large preview you intend to allow. Google also lists AMP as a route for its Discover image recommendations. Read the robots meta tag specification for directive scope and behavior.

4. Make the image discoverable and accessible

Metadata pointing to an image is useful only if Google can fetch that image. Google can discover images through the src attribute of standard HTML <img> elements; it does not index CSS background images. When using responsive image markup, provide an <img src> fallback.

<picture>
  <source
    srcset="https://example.com/images/article-hero-1280x720.avif"
    type="image/avif"
  >
  <img
    src="https://example.com/images/article-hero-1280x720.jpg"
    width="1280"
    height="720"
    alt="The subject shown in the article’s lead image"
  >
</picture>

Ensure the page and image URL are accessible to Googlebot. If an image might otherwise be difficult to discover, an image sitemap can expose its URL. See Google’s image discovery guidance.

5. Verify the rendered page, not just the CMS setting

1. Open the published page and inspect its rendered HTML head.
2. Confirm that og:image points to the intended absolute image URL.
3. If structured data also names an image, confirm that it uses the same intended image.
4. Check the image URL itself: it should load successfully and be accessible to Googlebot.
5. Look for page-level robots settings that suppress image previews or limit them to standard.
6. Check the actual crop at a 16:9 presentation and confirm that the important subject remains visible.

A CMS field can look correct while a theme, plugin, or page template emits a different value. Inspect the output for the specific URL you want indexed.

6. Common problems and fixes

Symptom	Likely cause	What to check
Discover shows a different image	Google selects previews automatically from available signals and page content.	Make the preferred image relevant and prominent, align og:image and structured data, and check image quality and crop. The signals influence selection but cannot guarantee it.
The preview is small	The page does not permit a large preview, or another robots directive limits it.	Allow max-image-preview:large and check page-level robots output.
The intended image is not considered	The image URL may be inaccessible or discoverable only through CSS.	Use a crawlable HTML <img src>, provide a fallback for responsive markup, and check access to both page and image.
The crop hides the subject	The source image is not composed for a landscape 16:9 presentation.	Recompose or crop the image so its subject stays visible in landscape form; keep the image representative of the page.
Different images appear across pages	A shared template, plugin, or stale CMS field may emit inconsistent page-level metadata.	Inspect the rendered output for each affected page and make its preferred image consistent with its content.
Metadata looks right but selection does not change	Metadata is a signal, not a command, and image selection is automated.	Verify the page output and image access, then allow Google to process the updated page. Do not treat the tag as an override.

7. Keep image capture and review practical

When reviewing many pages, prioritize the checks that catch implementation mistakes: the rendered image URL, access to that image, metadata consistency, preview permission, and the landscape crop. Avoid repeatedly fetching or recapturing identical pages if the content has not changed. Google’s documentation gives image eligibility and discovery guidance, not a guaranteed processing time or a promise of Discover traffic.

For a visual review, inspect the page as a visitor would see it at a wide viewport and at a mobile viewport. That can reveal overlays, consent banners, or responsive layout changes that obscure the content while you check whether the image represents the page. A screenshot helps with visual review; it does not change Google’s image-selection system or replace crawlable metadata.

Or skip the browser setup

If you need screenshots of the page to review its rendered layout, ScreenshotNeo can return a screenshot from one API request. This example requests a WebP image of the article page:

curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://example.com/article -o shot.webp

See the ScreenshotNeo API documentation for request options. ScreenshotNeo accepts cookie or consent banners before capture and removes more than 60 known consent platforms, newsletter popups, and chat widgets; each of those steps can be turned off. Bot checks, blank pages, timeouts, failed loads, and cache hits cost nothing, and response headers report the page verdict and billing status. Its MCP server lets AI agents use take_screenshot, get_page_info, and capture_pdf. The Free plan includes 1,000 screenshots per month with no card; paid plans start at $5 for 3,000 screenshots.

Sign up for 1,000 free screenshots a month with no card.

Frequently asked questions

Does setting og:image guarantee that Discover uses it?

No. Google’s image-preview selection is automated. Page metadata can influence it, but does not force the selected thumbnail.

Should I use og:image or schema.org?

Google recognizes both as image sources. Use the approach your publishing system can maintain accurately. If both are present, keep their image URLs consistent.

Does max-image-preview:large select my preferred image?

No. It permits a larger preview; it is separate from the signals that identify a preferred image.

What image dimensions should I use?

Google recommends at least 1200 pixels in width and more than 300,000 pixels total, composed for a 16:9 landscape presentation. Meeting those recommendations does not guarantee Discover inclusion or a particular image choice.

Can Google use a CSS background image as the page image?

Google says it does not index CSS background images. Provide a crawlable HTML <img src> when the image should be discoverable.