How does the LinkedIn algorithm work?

The mechanism is a staged test rather than a single decision. A new post goes to a fraction of your connections and followers. If those people stop, read, and respond, it is shown to a wider group, and the process repeats. If they scroll past, distribution quietly stops. This is why the first hour matters more than anything you do afterwards.

Not all responses count equally. A comment is worth more than a reaction because it costs the reader more, and a comment you reply to is worth more still, since it signals a conversation rather than a broadcast. Time spent on the post is also read as interest, which is why formats that hold attention are favoured.

Consistency of subject matters because the system is classifying you as well as the post. An account that keeps writing about one thing teaches the ranking system exactly who to run the first test on, and those readers are the ones most likely to respond. An account that changes subject every week resets that test audience each time, which reads in analytics as reach that never compounds.

Being present after publishing is distribution work, not politeness. Every reply you leave under your own post is a fresh engagement event while the post is still being tested, and a thread where the author is answering pulls other readers into commenting. The same post abandoned at publication and tended for an hour can finish with very different reach.

Most published advice about the algorithm is folklore: banned words, magic posting times, penalties for external links. The parts that survive scrutiny are unglamorous. Post about a consistent subject, to an audience that has a reason to care, in a format that holds attention, and be present when people reply.

What the ranking system cannot do is make people care. It is a distribution mechanism, not a demand generator, and it amplifies whatever response your writing earns. A post nobody wanted to read does not get rescued by timing.