LinkedIn's rebuilt feed: the new ranking signals

LinkedIn has quietly rewritten the rules of who sees what. Social Media Examiner reports that the platform has pushed through one of its largest feed overhauls to date, shifting ranking away from broad reach and toward relevance scored at the level of the individual viewer. The implication for B2B brands is sharper than the trade press suggests: the era of writing a post for "LinkedIn" is over. You are now writing for a named subset of your network, and the algorithm will decide which subset.

The mechanism matters. The feed used to reward early velocity: likes and comments in the first hour told the system to fan a post out widely. That logic is being retired. The new system weights topical authority (does this author consistently post about this subject), relationship strength (does the viewer actually interact with this person, not just follow them), and dwell-time signals (did the viewer stop, expand, read, return). Reach is no longer a popularity contest decided in the first hour. It is a relevance match decided per viewer, per scroll.

This punishes three habits that dominated executive LinkedIn for the past five years.

The first is range. Founders and senior marketers who pivot between leadership homilies on Monday, a product announcement on Wednesday, and a personal anecdote on Friday now look incoherent to the ranker. The system cannot decide what they are an authority on, so it discounts them across the board. Topical consistency, the thing every editor has nagged executives about for a decade, has finally been priced into distribution.

The second is engagement theatre. Pods, "great post!" comment trains, and the ritual tagging of five colleagues used to manufacture early velocity. Those signals are being downweighted in favour of behaviours the platform considers harder to fake: saves, shares into DMs, profile visits triggered by a post, and follows that come from a single piece of content. For a CMO measuring LinkedIn properly, this is welcome. The metrics the algorithm now rewards are the metrics that correlate with pipeline.

The third is the broadcast mindset. A post that lands with 200 of the right people, namely procurement leads at insurers, programme officers at multilaterals, plant managers at industrial groups, will now outperform a post that lands with 20,000 of the wrong ones. The feed is being tuned for the former and against the latter. Vanity reach is becoming structurally harder to achieve and structurally less valuable when you do.

What replaces it is something closer to a professional recommendation engine. The viewer's behaviour, what they read to the end, what they save, whose profile they revisit, is feeding back into who they see next. For an executive at HOLCIM or a director at a UN agency, this means the audience that matters most, the one already half-persuaded, is the audience the algorithm is most aggressively curating toward you. Provided you give it something to recognise you for.

Three operating shifts follow.

Pick a lane and stay in it for at least a quarter. Two or three adjacent themes, posted consistently, will out-distribute a wider repertoire. A head of sustainability at an industrial group should sound like a head of sustainability in every post, not occasionally a parent, occasionally a runner, occasionally a management thinker. Save the range for the "About" section.

Write for the save, not the like. A post that a regulator screenshots into a working group, or that a banker forwards to a client, is now sending the algorithm a far stronger signal than 400 likes from second-degree connections. The format implications are concrete: tighter arguments, named examples, numbers a reader can quote in a meeting. Documents and longer text posts that reward dwell time are being favoured over one-line provocations designed to bait quick reactions.

Treat the comment section as the second post. With relationship strength now weighted heavily, the people who comment on your posts are the people most likely to see your next one. Replying substantively, not with "thanks!", compounds. For institutions with a deep bench, this is a coordination problem worth solving: when five senior people at the same organisation reply thoughtfully to one another's posts, they are training the algorithm to bind their audiences together.

The losers in this rewrite are the executives who treated LinkedIn as a megaphone and the agencies that sold them reach. The winners are the operators who already understood that B2B authority is built narrow and deep, with the same hundred buyers, regulators and peers seeing you sound like yourself, on the same subjects, for years. LinkedIn's algorithm has finally caught up with how reputation actually works.