We checked the link previews of the top 10,000 websites: 42% have no og:image
You paste a link into Slack, LinkedIn or X and get a bare URL with no picture. We wanted to know how common that is among sites that should know better, so on 17 November 2026 we fetched the homepage of each of the top 1
You paste a link into Slack, LinkedIn or X and get a bare URL with no picture. We wanted to know how common that is among sites that should know better, so on 17 November 2026 we fetched the homepage of each of the top 10,000 domains on the Tranco list, read their Open Graph and Twitter Card tags, downloaded every og:image and measured it against the limits the platforms document.
5,927 homepages could be read. The rest blocked unknown bots (1,420), had no public website (1,317, mostly CDN and DNS domains), did not answer (692), disallowed us in robots.txt (269) or returned other errors. Every number below says what it is a share of.
Five numbers
| Finding | Share |
|---|---|
Homepages with no og:image
|
42.0% (2,489 of 5,927) |
og:image smaller than LinkedIn's 1200Γ627
|
38.0% (1,242 of 3,266 measured) |
| Homepages that pass Facebook, LinkedIn and X at once | 31.8% (1,883 of 5,927) |
Declared og:image that does not load as an image |
4.4% (150 of 3,438) |
Homepages with no twitter:card
|
50.6% (2,997 of 5,927) |
Being popular helps, but less than you would think: in the top 1,000 domains, 34.7% of the homepages we could read still had no og:image.
What goes wrong
No image at all. The single biggest failure. A shared link then shows the title and description with no picture, or a picture the platform picks itself.
Too small. LinkedIn documents 1200Γ627 as the minimum for a large preview; 38% of the images we could measure were smaller. 22% were even smaller than Facebook's 600Γ315, below which Facebook shows a small thumbnail instead of the large card. 7.8% were under 200Γ200, where Facebook shows no image.
Broken. 4.4% of declared images did not load as an image: an error status, an HTML page instead of a picture, or a dead host. Another 3.8% used a relative URL (/img/og.png). Browsers resolve that; link-preview crawlers are not required to.
Format. Some sites point og:image at an SVG or AVIF. X documents JPG, PNG, WebP and GIF only (no SVG), so stay with JPEG or PNG.
No twitter:card. X falls back to og:title, og:description and og:image when the twitter: versions are missing, but twitter:card has no Open Graph equivalent: without it, X may render the small summary card. 38.5% of homepages ask for summary_large_image.
The fix is four lines
<meta property="og:title" content="Your page title">
<meta property="og:description" content="One sentence about the page">
<meta property="og:image" content="https://example.com/og/home.png">
<meta name="twitter:card" content="summary_large_image">
One og:image tag, an absolute https:// URL, an image of 1200Γ630 pixels (JPEG or PNG under 5 MB is safe everywhere), and twitter:card set to summary_large_image. Then change the image URL (for example ?v=2) when you replace it, because every platform caches previews: X for about a week, Facebook until you click Scrape Again in its Sharing Debugger.
Method and data
- Homepages only, fetched once with a self-identifying user agent; robots.txt respected (RFC 9309); public addresses only.
- Tags read with a standards-compliant HTML parser (the first value of each tag counts); every
og:imagedownloaded and its pixel size read from the file. - Pages that add their tags with JavaScript after load were measured as served, which is what link-preview crawlers see too.
The per-domain CSV and the aggregate JSON are free under CC BY 4.0 on GitHub. The full write-up, with every table and the FAQ, is on shareglint.com.
Disclosure: I build Shareglint, an API that renders share images from HTML templates. The census used the parser behind its free OG Checker, which you can run on your own page without an account.
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes β full credit and traffic to the original publisher.