Skip to content

Wrong image in Google results: set a preferred thumbnail

Google showing the wrong image in search results? Declare a preferred image with primaryImageOfPage, the main entity's image or og:image, then check crawlability.

By Crawlito · 8 min read

In short

  • Google says its choice of image preview is completely automated, but you can influence it with schema.org primaryImageOfPage, an image on the page's main entity, or the og:image meta tag.
  • Google's best practices for that preferred image: relevant and representative of the page, not a generic image such as your site logo, no text in the image, no extreme aspect ratio, high resolution where possible.
  • Google finds images in the src attribute of img elements and does not index CSS images, so a hero set as a CSS background is not a documented candidate.
  • Google says crawling preferences for Googlebot-Image affect all Google Search features where images are presented, so a robots.txt rule that blocks your image folder can remove your best photo from the candidates.
  • max-image-preview:none tells Google not to show an image preview at all; standard or large set the maximum size, not which image is chosen.
On this page

If Google shows your logo, an ad banner or an unrelated photo next to your result, tell it which image you prefer. Google's image SEO documentation names three metadata sources for this: the schema.org primaryImageOfPage property, an image attached to the page's main entity, and the og:image meta tag. Then make sure that image is a real <img> in the HTML, that Googlebot-Image can fetch it, and that nothing in your robots rules hides it.

Set expectations first. Google says its selection of an image preview "is completely automated" and draws on a number of sources. Metadata lets you influence the choice. It does not let you dictate it, and no setting forces a particular thumbnail.

Why Google picks the wrong image

Google's documentation says the preview selection takes several sources into account, for example for a text result image or the preview image in Discover. It does not publish the full list or how they are weighed. What you can see from the outside is which images Google had to choose from. The table below lists common patterns to check; they are diagnostic guesses, not causes Google documents:

What you seeLikely cause on the page
Your logoThe logo is the only crawlable <img>, or your og:image or JSON-LD points at the logo
An ad or promo bannerThe banner is a normal <img> near the top, while the real photo is missing, blocked or a CSS background
A photo from another articleA related-posts widget or sidebar ships <img> tags, and the article's own image is not declared anywhere
An old imageThe metadata still names the previous file, or Google has not recrawled the page since you changed it
No image at allmax-image-preview:none is set, or no usable image was found

The common thread: Google can only pick from images it can find and fetch. Fix the candidates first, then declare a preference.

How to tell Google which image you prefer

Google lists three ways to declare a preferred image. You do not need all three, but they should agree.

Option 1: primaryImageOfPage on the WebPage

Schema.org defines primaryImageOfPage as the main image on the page, on the WebPage type. Schema.org lists ImageObject as the expected value, and Google's example uses a plain URL, which Google says it accepts as "a URL or ImageObject".

html
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "WebPage",
  "url": "https://www.example.com/guides/red-kettle",
  "primaryImageOfPage": {
    "@type": "ImageObject",
    "url": "https://www.example.com/images/red-kettle-1600.jpg"
  }
}
</script>

Option 2: an image on the main entity

If the page is about one thing, such as an article or a product, give that thing an image and connect it to the page with mainEntity or mainEntityOfPage. Google's example is a BlogPosting:

html
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "How to descale a kettle",
  "mainEntityOfPage": "https://www.example.com/guides/red-kettle",
  "image": "https://www.example.com/images/red-kettle-1600.jpg"
}
</script>

This is often the easiest fix, because many sites already output Article or Product markup. Check that its image is the hero photo and not a site-wide default.

Option 3: og:image

html
<meta property="og:image" content="https://www.example.com/images/red-kettle-1600.jpg">

The Open Graph protocol describes og:image as an image URL that should represent your object. Many CMS themes set it to the site logo on every page when a post has no featured image, which is exactly the generic image Google advises against. The share-card side of Open Graph, with its scrapers, sizes and caches, is covered in fixing Open Graph link previews.

Make the three sources agree

Google's page does not say which source wins if they disagree. Do not leave it to chance: point all of them at the same file, at the same absolute URL. For an image used on several pages, Google also recommends referencing it consistently with the same URL so it can cache and reuse it.

What makes a good preferred image

Google gives four best practices for the image you put in schema.org markup or og:image:

  1. Relevant and representative of the page. The photo of the product, the subject of the article, the dish on the recipe page.
  2. Not generic, and no text. Google names your site logo as an example of a generic image to avoid, and says to avoid an image with text.
  3. No extreme aspect ratio. Avoid images that are too narrow or overly wide, such as a thin banner strip.
  4. High resolution, if possible.

This section of Google's guide gives no pixel minimum. Image size limits for previews are set separately with robots rules; max-image-preview values and effects covers them, including the large previews used in Discover.

Two more points from the same Google page help the right image look like the obvious candidate: place images near relevant text, and use short, descriptive filenames (red-kettle-descaled.jpg, not IMG00023.JPG). Descriptive alt text helps too, as covered in writing image alt text.

Check that Google can find and fetch the image

Metadata pointing at an image Google cannot use is wasted. Work through these checks on the exact URL that shows the wrong thumbnail.

The image is an img element with a real src

Google says it finds images in the src attribute of an <img> element, including one inside <picture>, and that it does not index CSS images.

html
<!-- Google can find this -->
<img src="/images/red-kettle-1600.jpg" alt="Red enamel kettle on a gas hob">

<!-- Google does not index this -->
<div class="hero" style="background-image:url(/images/red-kettle-1600.jpg)"></div>

A full-width hero built as a CSS background can explain why the logo or a banner wins: if they are the only <img> elements, they are the only images Google finds in the markup. If you use srcset or <picture>, Google recommends always giving a fallback URL in src, because some browsers and crawlers do not understand those attributes. If a lazy-loading script keeps a placeholder in src and swaps in the real file later, inspect the rendered HTML in Search Console's URL Inspection tool to confirm the real URL is there.

The image URL is not blocked

Google's common crawlers page says crawling preferences addressed to Googlebot-Image affect Google Images, Discover, Video mode and all features in Google Search where images, logos and favicons are presented. A Disallow: /uploads/ or Disallow: /media/ rule meant to tidy up Google Images can therefore remove your best photo from the candidates. Images on a CDN host are governed by that host's own robots.txt. The Googlebot-Image token guide shows how the image crawler matches groups.

bash
# Find the declared image URLs in the raw HTML
curl -s https://www.example.com/guides/red-kettle \
  | grep -Eo '"primaryImageOfPage"[^}]*|"image" *: *"[^"]*"|<meta[^>]*og:image[^>]*>'

# Check the image itself
curl -sI https://www.example.com/images/red-kettle-1600.jpg

# Check robots.txt on the host that serves the image
curl -s https://www.example.com/robots.txt

You want a 200 with an image/ content type, and no Disallow rule in the group Googlebot-Image follows that matches the path. The grep is a quick screen; confirm odd results in the page source. Use a format Google supports in img src: BMP, GIF, JPEG, PNG, WebP, SVG or AVIF. If the image lives on a CDN and is missing from Google Images too, image sitemap checks for CDN images walks through the page, sitemap and file separately.

No robots rule suppresses the preview

The robots meta tag documentation has several rules worth checking:

RuleEffect on images in results
max-image-preview:noneNo image preview is shown
max-image-preview:standard or largeSets the maximum preview size; without the rule, Google may show a default-size preview
noimageindexImages on this page are not indexed
nosnippetBlocks text snippets, but Google says a static image thumbnail may still be visible when it results in a better user experience

None of these chooses the image. They only limit whether and how large it may appear. Look for them in both the HTML <meta name="robots"> tags and any X-Robots-Tag header.

Keep the wrong image out of the running

Declaring a preference is the main fix, but you can also remove the bad candidates.

  • Logos. Keep the logo in the header as it is, and make sure no metadata on article or product pages points to it. Replace any site-wide default og:image with the page's own image, and fall back to a representative photo, not the logo.
  • Self-hosted promo banners. If the banners are your own files in a dedicated folder, a rule for Googlebot-Image stops Google's image crawler from fetching them. Google's image removal guide gives the pattern:
User-agent: Googlebot-Image
Disallow: /images/promo/

Scope this tightly. Only block files you never want shown, and note that Google's crawlers follow only the most specific group that matches them, so a Googlebot-Image group replaces your * rules for that crawler; copy across any other rules it should follow. The same guide also offers a noindex X-Robots-Tag header on the image files as an alternative; in that case the image URLs must stay crawlable so Google can read the header.

  • Third-party ads and widgets. You cannot set robots rules on another company's ad server. Your levers are the metadata above and making sure your real image is a crawlable <img> near the main text.

After the fix: recrawl and wait

Google has to recrawl and reprocess the page before a new thumbnail can appear. Confirm everything with curl first, then use Request indexing in Search Console's URL Inspection tool for the most important URLs. Google does not publish how long a thumbnail change takes, and because selection is automated, a correct setup is not a guarantee that your preferred image is shown. Keep the image URL stable once it is right.

Frequently asked questions

Can I force Google to use a specific thumbnail?

No. Google says image preview selection is completely automated. You can influence it with primaryImageOfPage, the main entity's image or og:image, and by making sure the image is crawlable.

Why does Google show my logo instead of the article image?

Usually because the logo is the most visible crawlable image or your metadata points to it. Google advises against generic images such as a site logo as the preferred image, so declare the article's own image and check that it is an <img> with a real src.

Does Google use og:image for search result thumbnails?

Google lists og:image as one of the metadata sources you can use to indicate a preferred image. It is one input to an automated choice, not a guaranteed override.

Will a CSS background image ever be used as the thumbnail?

Google says it does not index CSS images, so a background image is not a documented candidate. Use an <img> element, or a <picture> element with an img fallback.

Does max-image-preview:large make Google pick a better image?

No. It sets the maximum size of the preview Google may show. Which image is shown depends on the automated selection and the metadata you provide.

Check what crawlers receive

Many wrong thumbnails trace back to what the raw HTML contains: missing metadata, a CSS hero, or a blocked image path. The free check reads one page the way crawlers do, with no account needed, including robots rules and whether key content is in the raw HTML. For template-level problems across many pages, an audit reviews up to 500 pages with evidence and prioritized fixes.

Filed under Structure and schema. Topics: images, google, metadata, structured-data, troubleshooting