Open Graph Preview

The tags are read with the browser's own HTML parser, which is the definition of what the markup means. That matters for two things a regular expression gets wrong. It resolves character references, so a title written as The 24" Monitor Review & Guide comes back as the sentence a crawler reads rather than as the escape sequence. And it shows what happens when a value broke out of its attribute, which is exactly the failure a hand-written generator produces for a title with a quote in it: the tag looks present in the source and the parser finds a truncated value or none at all.

Seven platforms are previewed and six of them cut the title at a different length. A 200 character title becomes 60 characters in Google, 70 on X, 88 on Facebook, 119 on LinkedIn, 120 in Slack, and stays whole in Discord. That cut is applied to the text rather than hidden with a CSS clamp, so what you see is the string the platform gets, ending where it ends. Every number is shown with the documentation it comes from, and the page says plainly that these are documented rather than measured: nothing running in a browser can see what Facebook renders. What is measured is the cut itself.

The image is measured too. It is fetched from this page and its real width and height are read, so the 200 by 200 minimum Facebook enforces, the 1200 by 630 recommendation and the 1.91:1 crop are checked against the file rather than against what you believe about it. A square image is reported as losing its sides; a 4:1 banner as losing its top and bottom. The commonest failure of all is checked first: og:image has to be an absolute URL with a scheme, because a crawler has no page to resolve a relative one against, and a relative one produces a share with no picture and no error anywhere.

The checks come from the Open Graph protocol. The four required properties are og:title, og:type, og:image and og:url, and the twelve object types are named, so an og:type of "blog" or "product", both of which are common and neither of which is in the protocol, is reported as an error rather than passed through. A missing og:url is explained rather than just flagged: without it a share resolves to whatever URL the sharer was on, so one page shared from a tracking link, a search result and a redirect becomes three objects with three separate counts.

The tags that produce the preview are shown and can be copied, with every value escaped for an attribute. They come from the same module the Meta Tag Generator uses, so the two tools cannot disagree about what a tag means or about where a title gets cut, and the generator's output can be pasted straight back in here and read as the same values. That round trip is one of the tests.