What the LinkedIn algorithm is
When you open LinkedIn, the feed is not a chronological list. Models decide which posts deserve to be shown, in what order, and to whom. What LinkedIn documents comes down to two stages: retrieval, which selects candidate posts, then ranking, which determines what the member actually sees. In the engineering team's own words, retrieval decides which posts reach the ranking stage, and ranking decides what is displayed.
Source: "Engineering the next generation of LinkedIn's Feed", LinkedIn Engineering blog, March 12, 2026.
What is documented, and what is not
The same engineering article describes the signals in play. On one side, profile information such as industry, experience, skills and geography. On the other, engagement history: what you read, liked, commented on, revisited or simply scrolled past. LinkedIn states that ranking relies on professional signals and engagement habits, not demographic attributes.
- Documented: the feed is personalized, it combines profile and interaction history, and it predicts passive actions (click, skip, long read) as well as active ones (like, comment, share).
- Documented: post and profile representations are refreshed within minutes, so freshness matters.
- Not documented: the exact weight of each signal and the scoring formula. That LinkedIn page does not give them.
Anything circulating elsewhere about ideal posting times, magic lengths or numeric penalties for external links is observation or rumor, not documentation. Treat those tips as hypotheses to test on your own posts.
What it changes for your posts
Because the feed predicts interest from past interactions, content relevant to a specific audience is more likely to rank well than generic content. Write for identified readers, with a first line that makes them want to read on: see our LinkedIn hooks and, for example, these post templates for an accountant.
Early reactions are engagement signals like any other, which is why some creators organize into mutual-support groups. That practice has limits and risks for your account, detailed in our article on how LinkedIn engagement pods work and their limits. To go further, browse the LinkedIn algorithm category of the blog, or the LinkedIn glossary.
In practice, three habits follow from what LinkedIn documents. First, pick one audience per post, since the feed matches content to members whose history shows an interest in it. Second, aim for reactions that show real attention, such as comments and long reads, because the system predicts both passive and active actions. Third, publish regularly and check your own results, since LinkedIn gives no public formula and the only reliable feedback is what your posts actually do.
If you want genuine reactions from active members on your posts, you can create a Podavera account: 20 credits are offered at sign-up.