Business of GTM

LinkedIn Profile Relevance Sends Posts Past Followers After One Outside Reaction

LinkedIn says the LLM system complements network signals and recommends a post only after a like, comment or share from an unconnected member. Its online tests lifted Feed Updates Viewed 0.54%.

Black-and-white halftone photo of LinkedIn CEO Dan Shapero speaking into a microphone, in front of a red circle on a beige background.
Dan Shapero, CEO of LinkedIn, speaking in January 2025, when he was its chief operating officer. Photo illustration by Business of GTM. Original photo courtesy of LinkedIn.

TL;DR

  • LinkedIn launched Profile Relevance on November 9, using LLM embeddings to match members' profiles against posts from outside their network.
  • It adds a second route past the follower network, which opens only after one reaction from an unconnected member.
  • LinkedIn's own tests show lifts under 1% on feed views, with the largest gains among readers who have little activity history.

LinkedIn has a new way to show posts to people who don't follow their authors, and it waits for a stranger to react first.

In a November 9 engineering post, LinkedIn described Profile Relevance. The system uses large language model embeddings to match a member's profile against posts from outside that member's network. LinkedIn calls it a key source of out-of-network posts. It says the system complements network and engagement signals instead of replacing them.

Profile Relevance is a second route past an author's follower network, and a post can only take it after someone outside that network reacts. Follows still decide who sees a post first.

One outside reaction opens the route

Posts recommended through Profile Relevance need "a single viral action (like, comment, or share) from an unconnected member," according to LinkedIn's post, written by Shiyong Lin. A reaction from a connection doesn't meet that threshold.

The author's network sees the post through LinkedIn's existing sources. One person outside the network likes, comments or shares. Only then can Profile Relevance match the post to members whose profiles fit.

Koka Sexton, a marketing leader who sells LinkedIn profile and growth services, describes three years of past work at LinkedIn. In a post on the change, Sexton wrote, "Engagement is a gate, not a score." Sexton argues the gate leaves engagement pods with nothing to do on this route, because pod members are connected to the author.

The match reads the reader's profile

On the member side, Profile Relevance embeds a reader's headline, summary, skills, work history and education. LinkedIn says it leaves out name and gender. The post side is built from the post's text, titles and transcripts, plus author metadata such as location.

Sexton reads this as a reason for creators to rewrite their own profiles. The Profile Relevance post doesn't say it reads the author's profile. LinkedIn's March 2026 feed post does put author name, headline, company and industry into post prompts. Those prompts belong to its main retrieval system.

For a B2B author, the post is what gets matched against a buyer's skills and work history. A post that names the work a RevOps lead or a CRO does likely gives the model more to match than a post about the author. Audience size still matters for the first step, getting the outside reaction.

The March rebuild already moved this way

LinkedIn's March 12 rebuild replaced several retrieval sources with a single LLM-embedding system. That system balances posts from a member's network, from accounts the member follows and suggested posts. A 2025 paper by LinkedIn engineers had already run LLM retrieval of suggested posts in an online test, with significant gains among newer members who lack strong connections. LinkedIn has also tested a For You and Network feed split since September.

The Profile Relevance results come from LinkedIn's own online tests, measured against its previous source, MemberEBR:

  • Feed Updates Viewed rose 0.54%.
  • Viral rate rose 0.94%.
  • Hide rate fell 18.8%.
  • Among readers with little activity history, viral engagement rose 17.7% and daily active users 1.8%.

The lifts on views and viral rate are under 1%, and the largest gains went to readers LinkedIn knows little about. Nothing in the post measures what happens to an author's reach. The 18.8% drop in hides suggests the profile match picks posts members don't mind seeing.

A stranger's first like now stands between a post and its buyers

Profile Relevance covers posts from people and companies a member doesn't follow. The same threshold applies to a company page and a founder's post. LinkedIn is widening reach past the network in steps it describes as additions, and each step ties distribution to the reader's profile and to signals the author doesn't control.

Engagement is a gate, not a score.

Koka Sexton, marketing leader who says he spent three years at LinkedIn

LinkedIn's November 9 post reports a 0.54% lift in Feed Updates Viewed and a 0.94% lift in viral rate, and nothing on what happens to an author's reach. A reach figure for authors would settle whether the outside-reaction route helps them.

By Business of GTM