When auto-generating a post such as a link post, how does SocialChorus determine the image, title, and description from a link? How do the social networks such as Facebook, Twitter, and LinkedIn determine the image, title, description, and cropping?
SocialChorus leverages a tool called embed.ly, which follows the industry standards. The tool looks for Open Graph tags such as og:image, og:title, and og:description and will also scrape the website page for possibly relevant content using internal heuristics.
The following URL can be used to test the results of embed.ly for a specific link (replace <link> with the link that you are testing):
The og: metadata tags were first developed by Facebook and have become an industry standard. The standards for og: tags can be reviewed on Open Graph’s website. To review the use of og: tags by some of the major social media networks:
+ Facebook looks first for Open Graph tags, then falls back on internal heuristics - Facebook Developer Documentation
+ Twitter looks first for Twitter card tags, then looks for Open Graph tags - Twitter Developer Documentation
+ LinkedIn also looks for Open Graph tags - LinkedIn Developer Documentation
Please note that while we are able to describe the metadata tags, we are not able to define the scraping methodology for any of the social networks or even embed.ly - these algorithms change over time, and are not publicly disclosed.
Implementing Open Graph tags in a website's metadata can lead to more consistent results when posting content to SocialChorus, Facebook, Twitter, and LinkedIn, as they all check for og: tags before scraping the page. However, sometimes, different sites will make different decisions even about tagged content. For example, SocialChorus has different image size requirements for content cards than Facebook has for wall posts. If an og:image is too small, embed.ly may search the website page for a large enough image file.
Also, occasionally content is not accessible when a link is previewed - for example, some websites require that a user is signed in to see the content, some links lead to redirects rather than content, and occasionally a page that looks like a website is actually a Java plug-in, all of which can lead to a blank link post.