LinkedIn has explained how its feed picks posts to show people who do not follow you. LinkedIn out-of-network reach rests on a profile match. A model called Profile Relevance reads each member’s own profile and condenses it into a short summary of what they do. Candidate posts are then scored against that summary. The trade press covered the engineering write-up on 9 November 2026, a week after LinkedIn detailed the analytics that show you the result.
Profile Relevance works from text. LinkedIn takes a member’s public profile and posting history and condenses them into a prompt-like summary. That summary becomes an embedding, and so does every candidate post. The score is how closely the two sit together on topic, location, job title, company, industry, seniority and education.
LinkedIn says it reviewed the output by hand before letting the model show members posts from people they do not follow. The engineering write-up describes two variants. In the Rewrite version, a large model pulls topics out of the text and the result is embedded. In the Non-Rewrite version, the raw text is embedded and more context survives. Rewrite lost geographic signal, so LinkedIn appended country names to the input to get it back.
The part that matters is whose profile is being read. It is the reader’s, not yours. Your follower count is not in the match. Neither is the group of colleagues who like everything you post within the first hour.
Three numbers came out of LinkedIn’s online test. Feed updates viewed rose 0.54 percent. The viral rate rose 0.94 percent. The hide rate fell 18.8 percent, and that is the only one of the three that moved meaningfully.
So the model is a relevance filter, built to stop showing people posts they do not want. It is not a distribution engine built to hand your posts to more strangers. If you were expecting a reach windfall, 0.54 percent is the number to sit with. That is most of what the platforms reward right now: better targeting of the attention that already exists, rather than more of it.
The gains were not spread evenly, though. Members with limited activity history, the group LinkedIn calls cold-start, did much better. Their viral engagement rose 17.7 percent and their daily active use rose 1.8 percent.
LinkedIn’s own research paper on the feed retrieval system reports the same skew. Twenty-four of its researchers published it in October 2025. Most of the platform-wide impact, they found, came from infrequent members and members with fewer connections. For that group, daily unique professional interactions rose 1.17 percent and revenue rose 3.29 percent.
So the extra distribution lands on people who post rarely and connect rarely. That is the quiet reader in your category who never comments. Write for somebody with no history with you, because the system is matching on their profile rather than on anything they have done.
Sam Corrao Clanon, LinkedIn’s director of creator products, put this plainly when the Reach metric launched in June. Posts “perform best in network when they’re about you and the things that you’re doing”, he told Forbes. More generally applicable expertise tends to travel further out of network.
There is now a mechanism underneath that observation. A post about your promotion matches nobody’s professional context except the people who already know you. A post about a costing problem specific to independent agencies matches every profile that names agency work. The match is literal, so the post has to be literal.
In practice, that means naming the industry, the role and the problem in the body of the post. Not in the hashtags underneath it. It also means the original content that travels is the kind somebody outside your circle can act on without knowing who you are.
LinkedIn detailed a set of creator analytics on 3 November 2026. The headline addition is a Post Performance Table. It lists every post with its metrics side by side. You can now compare impressions, views and watch time “without needing to open the analytics for each one of your posts individually”. Three smaller additions came with it. Followers gained shows which specific members followed from a post, and saves, sends and the in and out of network split are all there too.
The split itself is the Reach metric. It has sat in the Discovery section of post analytics since June. In-network is defined as your connections and followers. Out-of-network is everybody else, reached through feed recommendations, reshares and search. It is the one number that tells you whether the profile match is working. That makes it a better thing to measure than a raw impression count.
Nothing in the write-up says engagement signals were removed. Profile Relevance complements what LinkedIn already uses, including likes, comments and how close somebody sits to you in the network. A post nobody engages with still goes nowhere.
LinkedIn has also not published what share of a typical feed comes from outside the network. The research paper describes no separate out-of-network pipeline either. There is one retrieval system, and out-of-network candidates are simply the ones that matched on topic rather than on connection.
The match is scored against the reader’s summary, not yours, so rewriting your own headline does nothing for it. Your profile still matters for the people who land on it. That is where professional authority gets established. It is just not the lever in this particular system.
Yes. Profile Relevance complements LinkedIn’s existing signals rather than replacing them. Likes, comments and network proximity are all still in the ranking. What has changed is the documented route to people who have never seen you. That route does not run through engagement at all.
Probably not by much. Feed updates viewed rose 0.54 percent in LinkedIn’s test, and the viral rate rose 0.94 percent. Both are close to a rounding error. The 18.8 percent fall in the hide rate is the real result, and it means fewer irrelevant posts rather than more impressions.
Not for this reason. The summary that drives the match is built from the reader’s profile. Your headline does not decide whether a stranger sees your post. Write your profile for the people who click through after seeing it.
No, it depends what the post is for. A hiring announcement or a client win is meant for the people who already know you. A high out-of-network share on one of those means it reached the wrong audience. Judge the split against the job of the post.
The practical change is small and specific. Say who your post is for, inside the post, and then read the Reach split instead of the impression count. LinkedIn out-of-network reach is now a number you can check rather than a thing you assume.
The reach itself has not got easier. It has got legible, and that is worth the ten minutes it takes to look.
This is the kind of thing we work on with clients at Social Lady. If you want a second opinion on how your LinkedIn posts are reaching people, get in touch.
