Content from data started as a way to publish articles: a new row in a table or a new item in a feed becomes a page on your site. The same data is often worth a post. A new monthly figure, a product release in your changelog feed, a section of a long report: the company account should say something about it, and it usually doesn't, because somebody has to write it.

Content loops can now post to your company's X account and your LinkedIn Company Page. If X Organic or LinkedIn Organic is connected to the workspace, both show up as destinations in the Content from data wizard.

How it works

You set up the loop the same way as one that publishes articles: a source, a destination, a brief and a schedule.

Source. A BigQuery table, an RSS or Atom feed, a long document in your context tree, or your best-performing pages from Google Search Console. Each run takes the next item the loop hasn't covered yet.
Destination. "X post (from the connected account)" or "LinkedIn Company Page post". For LinkedIn you pick the Page from a list of the Pages the connected account can post to, meaning those where it is a Super admin or Content admin.
Brief. The content type, a writing style (your brand voice by default, or the LinkedIn thought-leadership style) and any instructions, such as which page to link or what never to mention.
Schedule. Hourly at most. A new loop starts paused.

Written for the network, not cut down from an article

A post gets its own rules instead of the search rules an article follows:

On X: one idea per post, within 280 characters with a link counted as 23, leading with the fact itself. At most one link and one hashtag.
On LinkedIn: the point goes in the first two lines, because LinkedIn hides the rest behind "see more". Short paragraphs, written as the company, at most three hashtags at the end. A link goes in as a link preview rather than pasted into the text.
On both: every number, name and date comes from the source item or from data the writer read during the run. No engagement bait, no clickbait, no emoji walls. Longer posts go through the same humanizer check as articles.

Before it posts, the writer reads the account's or Page's latest posts. If one already covers the same item, it posts nothing and says so. It also matches how your recent posts read: their length, emoji and hashtags.

Safe to leave running

Locked to one Page. A LinkedIn loop can only post as the Page you picked. Cogny refuses any other author, so the loop never posts from someone's personal profile. On X it posts as the connected account, so connect the company's account, not a person's.

Test first. When you create the loop, Cogny writes one test post from the newest item, with its character count, and posts nothing. The draft lands on the loop's page. Change the brief and press Test piece until it reads right, then turn the schedule on.

No draft step on the network. X and LinkedIn publish a post the moment it is sent, so the test post is your review. After that, every post is a ticket with its source item, the text that went out and the post's URL.

Credits. Writing a post uses credits like any loop run, and each X post also uses credits because X charges per post. A run with nothing new in the source posts nothing and costs nothing.

Who can use it

Content from data is part of Cogny Cloud. Connect X Organic or LinkedIn Organic under Settings → Integrations, then open Loops → Content from data.