Open Graph Checker

What Open Graph tags do, and why the preview is copy someone should review.

A shared link is often the first thing a reader sees of a page, and it is built from tags most sites never look at. Here is what those tags are, which ones matter, how platforms cache them, and the mistakes that leave a link sharing with no image or the wrong words.

What Open Graph is

The Open Graph protocol is a set of <meta> tags that describe a web page as an object: what it is called, what kind of thing it is, which image represents it and what its permanent address is. Facebook introduced it in 2010 so that any page could appear in its feed as a tidy card rather than a bare URL, and published the spec at ogp.me. It spread well beyond Facebook. LinkedIn, Slack, Discord, WhatsApp, iMessage and most chat and social apps build their link previews from the same tags, which is why the preview for one URL tends to look alike wherever it is pasted.

The tags live in the page's <head>. Each one names a property in a property attribute and gives its value in content:

<!-- the four the protocol requires -->
<meta property="og:title" content="How to Read a K-1">
<meta property="og:type" content="article">
<meta property="og:image" content="https://yourdomain.com/img/k1.png">
<meta property="og:url" content="https://yourdomain.com/blog/k1/">

<!-- optional, and used by nearly every card -->
<meta property="og:description" content="What each box means.">
<meta property="og:site_name" content="Your Firm">

A platform's scraper fetches the page when a link is first shared, reads these tags from the served HTML and stores what it found. It does not run the page's JavaScript in any reliable way, so a tag injected by a script after load may as well not exist. That is also why this checker reads the served HTML rather than the rendered page.

The six tags this checker reads

ogp.me lists four properties as required for every page and a longer list as optional. The checker reads all four required ones and the two optional ones that nearly every card displays:

Other properties are worth knowing even though they are not scored. og:image:alt describes the image for people who cannot see it, and ogp.me says a page that specifies an og:image should specify it too. og:image:width and og:image:height let a scraper lay out the card before the image has downloaded. og:locale declares the language. The checker lists any of these it finds under "other og: tags", so you can see what the page carries beyond the core six.

OG image size and format

Each platform publishes its own image guidance, and it changes, so the platform's own page is the source to trust. Facebook's current sharing documentation asks for images of at least 1200 x 630 pixels for high-resolution screens, an aspect ratio as close to 1.91:1 as possible so the image is not cropped in the feed, a floor of 600 x 315 pixels for the large link card, and a file no bigger than 8 MB. An image that meets those numbers generally works well on LinkedIn and Slack too, but check their documentation before you rely on it.

Two rules hold everywhere. The URL has to be absolute, starting with https://, because the scraper reads the tag with no knowledge of the page it came from, so a relative path such as /img/card.png resolves to nothing; this checker fails a relative image for that reason. And the image has to be publicly fetchable: an image behind a login, a bot wall or a robots.txt rule is invisible to the scraper even when it loads fine in your browser. The checker does not fetch the image, so it cannot tell you either of those things about the file itself. The OG Image Checker is the tool for the image and the card it produces.

Why the old preview keeps coming back

Platforms do not scrape a page every time it is shared. They scrape it once, store the title, description and image, and reuse them. Fix a missing og:image today and a link you share an hour from now can still appear without one, because the platform is showing you its stored copy. This is the single most common reason people believe their tags are broken after they have fixed them.

The remedy is to ask the platform to scrape again. Facebook's Sharing Debugger shows exactly what Facebook read from a URL and has a button to scrape it again. LinkedIn's Post Inspector does the same for LinkedIn. Chat apps rarely expose a tool; a changed query string on the URL usually forces a fresh fetch there. Run this checker first to confirm the tags in the served HTML are right, then refresh each platform's copy.

Do Open Graph tags matter for search?

They are not a ranking signal that anyone has documented, and a page that claims otherwise is guessing. Google does read two of them for display, though. Its title-link documentation lists content in og:title tags among the sources Google may use to write the blue link in a result, alongside the <title> element and the page's headings. Its site-name documentation says the system that picks the name shown beside a result considers og:site_name on the home page. So a careless og:title can surface in Google as well as on LinkedIn, and a missing og:site_name removes one of the inputs Google uses to name your site.

The preview is copy that travels alone

A link preview leaves the page behind. When an advisor, attorney or practice shares an article on LinkedIn, the card carries the og:title, the og:description and the image, and none of the disclosures, qualifications or context around them on the page. If the description was filled in automatically from the first paragraph, or copied from a meta description written to win clicks, a performance claim or an outcome can end up standing on its own in a colleague's feed, with the firm's name above it.

For firms whose marketing is reviewed before it goes out, that makes the preview text marketing copy in its own right. Write the og:title and og:description on purpose, read them as a reader of the card would, with no page around them, and review them with the page rather than leaving them to a plugin's defaults. Choose the image the same way. This checker shows every value in one place, which makes the text easy to read over; it flags what is missing, not whether the words are appropriate, and that judgement stays with whoever reviews your marketing. None of the ranking pages we reviewed for this checker raise it.

Common mistakes

• No og:image at all. The most common gap and the most visible one: the card collapses to a line of text. Pick a default share image for the site, then a specific one for the pages people share most.
• A relative image path. /img/card.png works in the page and fails in the scraper. Always the full https:// address.
• The home page's og:url on every page. A template that hard-codes the site root into og:url tells platforms every page is the home page. Shares can pick up the home page's card, and share counts pool on the wrong URL. The checker compares og:url with the page it fetched and warns when they differ.
• Two plugins, two sets of tags. A theme and an SEO plugin both write Open Graph tags and the page ends up with two different og:title values. Platforms pick one, not always the same one. This checker reads the first copy of each tag, so view the page source if the values look wrong.
• name= where property= belongs. The spec uses property. Many scrapers accept name as well, and so does this checker, but property is the form every platform documents.
• Tags injected by JavaScript. A single-page app that sets its meta tags in the browser serves the scraper an empty head. The fix is to render the tags on the server.
• Changing the tags and judging the result by a re-share. The platform's stored copy wins until you ask it to scrape again. Check the served HTML here, then use the platform's debugger.

What a good set looks like

On every page: an og:title written for the card rather than copied from a title tag with a site-name suffix, a one-sentence og:description that reads well with nothing around it, an absolute og:image at the platform's recommended size with an og:image:alt beside it, an og:url that names the page itself, the right og:type, and the site's name in og:site_name. X reads Open Graph values when its own tags are missing, but the card type is set only by twitter:card; the Twitter Card Validator audits those. Then check the page again after every template or plugin change, because Open Graph tags are generated in one place and break everywhere at once.