All In
FR EN
Back to blog
Algorithme & stratégie LinkedIn 17 min read

The LinkedIn Algorithm Going Into 2027: What's Confirmed

| By Patrick de Carvalho

Contents

Richard van der Blom has published an Algorithm Insights Report every year since 2019, and it has become the closest thing LinkedIn has to an audited annual statement: a Dutch researcher, an independent methodology, a sample in the millions of posts, no LinkedIn sponsorship attached. He has trained more than 350,000 professionals on how the feed actually behaves. The edition people are waiting for now is the one that would cover a full calendar year under 360Brew, LinkedIn's semantic ranking model, from January through December 2026. That edition is not out yet. I checked his site directly before writing a word of this piece, and I'll show you exactly what I found. This is not a summary of a report that doesn't exist. It's an account of what's actually confirmed, what's trending mid-cycle, and where the honest answer is "wait for the data."

In short: The van der Blom report covering a full year of LinkedIn's 360Brew algorithm is expected in the opening months of 2027, following his cadence since 2019, but it had not been published or scheduled with a date as of this writing. The last confirmed edition, covering 2024 data across 1.8 million posts, found organic reach down 50%, engagement down 25%, and follower growth down 59% year over year. Mid-2026 trackers (Dataslayer, PostUnreel, Project Neon) show the same order of magnitude continuing, with carousels at roughly 6.6-7% engagement against 2% for text-only posts. Treat any specific 2027 percentage you see quoted online as unverified until it links to a dated, published source.

What the last confirmed edition actually said

Start with what's actually on the record, because it's more useful than most people give it credit for.

Van der Blom's report covering 2024 (published in his 2025 edition) analyzed 1.8 million LinkedIn posts. Three numbers from that report have been quoted so often they've become background noise, which is a shame, because they're the baseline everything since has been measured against: organic reach down 50% year over year, engagement down 25%, and follower growth down 59%.

Metric 2024 data (2025 report) What it measures
Organic reach -50% vs. 2023 Views per post without paid promotion
Engagement rate -25% vs. 2023 Reactions, comments, and shares relative to reach
Follower growth -59% vs. 2023 Net new followers per active publisher
Sample size 1.8 million posts Independent, cross-industry, self-funded

None of that is a 2026 or 2027 number. It's the last full-year dataset anyone has published with a named methodology behind it. If you've seen a chart online claiming "LinkedIn reach in 2027," ask what report it cites. In most cases it cites nothing, or it cites this same 2024 dataset relabeled.

What matters for a company publishing today is the direction, not the decimal. Reach contracting while follower growth contracts even faster tells you the platform stopped rewarding accumulation and started rewarding relevance. That shift predates 360Brew. 360Brew accelerated it, and it sits inside the wider shift All In tracked in its LinkedIn B2B trends for 2026 and 2027 roundup.

I went looking for the 2027 edition. Here's what I found instead

This is the part most articles about "LinkedIn in 2027" skip, and it's the part I think matters most.

I went to richardvanderblom.com directly to check whether the next edition, the one covering a full year under 360Brew, had shipped. It hadn't. What the site currently sells as its flagship product is labeled "Algorithm Report 2026," alongside a corporate licensing option described as "Full Algorithm Report 2026, all chapters." Nowhere on the site is there a mention of a 2027 edition, a publication date for one, or a preview of its contents.

Two things follow from that, and I want to be precise about both.

First, the annual cadence itself is well documented. Van der Blom's site describes an audience that anticipates his report every year "because it has become the de facto industry standard for what actually works." Nine editions since 2019 is a pattern I'm comfortable extrapolating from, in the same conditional way you'd forecast a company's next earnings call from eight prior ones. If that cadence holds, the edition covering full 2026 data, a complete calendar year under 360Brew, should land in the first months of 2027.

Second, and this is the part I won't fudge: I don't know what data year the current "Algorithm Report 2026" product actually covers, because it sits behind a paywall and I have not purchased it. Given his naming convention in prior years, where the "2025 report" covered 2024 data, a "2026" label most plausibly covers 2025 data rather than a full year of 360Brew. But plausible is not confirmed, and I'm not going to cite figures from a product I haven't read. That would be exactly the kind of unsourced claim this piece exists to argue against.

So here is the honest state of things: the report you're waiting for to settle the 360Brew question with a full year of evidence does not exist yet, publicly, with a date attached. Everything below this point is either the last confirmed full-year data, or interim tracking published by named, dated sources through 2026, clearly marked as interim.

360Brew, one year into semantic ranking

360Brew is real and it is documented, which puts it in a different category from the report question above. LinkedIn's engineering team built it as a roughly 150-billion-parameter model that replaced the platform's older engagement-counting ranker. Instead of tallying likes, comments, and shares as independent signals, it reads content and evaluates it in context.

Third-party analysis published in March 2026 by UpGrowth, cross-referencing LinkedIn's own engineering material with independent tracking data, describes the system working across four alignment checks: profile coherence, network relevance, engagement patterns, and content consistency. In plain terms, the model checks whether what you wrote matches who your profile says you are, whether the people it's showing your post to actually care about the topic, and whether your recent posting history backs up the claim you're making today.

The practical result is what analysts have started calling an expertise mismatch penalty. A sales director who posts consistently about B2B pipeline management gets wider distribution on a pipeline post than an identical post from an account that jumped there from three unrelated topics last month. Thomas Pettauer, writing on January 26, 2026, gives the sharpest illustration: a profile labeled "Graphic Designer" posting about crypto trading gets its reach suppressed for the mismatch, while a "RevOps Director" posting about Salesforce, a topic that matches the title, gets the wider distribution. That's a genuine shift from 2023, when a well-timed post from almost any account could catch a wave regardless of history.

For a founder or executive at a small or mid-size company, this cuts in your favor more often than not. You likely have a narrower, more consistent expertise footprint than a large company's marketing account juggling five product lines. The system built to reward topical consistency is, structurally, easier to satisfy from a position of genuine specialization than from a position of trying to cover everything.

What the mid-cycle trackers show while we wait

Absent a full-year report, the best available evidence through 2026 comes from third-party trackers who pull data continuously rather than annually. None of them replace van der Blom's sample size or methodology. All of them are dated and named, which is the minimum bar for using a number in this piece.

Dataslayer published a tracking update on February 12, 2026, later updated in July 2026, that found document posts (carousels) holding a 6.60% engagement rate, the highest of any format it measures, against roughly 2% for text-only posts. The same update found posts containing an external link losing approximately 60% of the reach an identical link-free post would get. Notably, the update also reported overall 2026 performance continuing to track the same direction as the 2024 data: views down roughly 50%, engagement down roughly 25% against the prior comparison period. That's the old finding holding steady rather than a new one, which is itself informative: nothing in the mid-2026 tracking data contradicts the last confirmed report.

PostUnreel's 2026 tracking put document-post engagement closer to 7%, in the same range as Dataslayer's number, which is the kind of independent convergence that makes a mid-cycle figure worth citing even without an annual audit behind it.

Project Neon's January 2026 analysis focused on timing rather than format: posts that kept collecting saves and substantive comments between 24 and 72 hours after publishing outperformed posts that got an immediate burst of shallow engagement and then went quiet, by a factor of four to six. That lines up with what 360Brew is built to do: reward sustained relevance signals over a quick spike.

Formats: carousel, text, and video

Put the interim numbers next to each other and a format hierarchy is holding steady through 2026.

Carousels, native PDF documents uploaded directly to LinkedIn, keep the top spot on engagement and, more importantly, on dwell time: readers spend two to three minutes with a well-built carousel against fifteen to thirty seconds on a typical text post, according to the last full van der Blom dataset. Dwell time is one of the clearest signals 360Brew has to work with, because it doesn't require the reader to do anything except keep looking.

Text posts, the 700-to-900-character range that still makes up the majority of LinkedIn's published volume, remain the right tool for a quick take, a reaction to industry news, or a short field observation. They will not out-engage a carousel, and they were never meant to.

Video, up sharply in volume since 2023 according to the last confirmed report, performs respectably in short vertical form but continues to trail carousels on engagement. For a leadership team without in-house production capacity, video is the format to use selectively, for a client testimonial or a product walkthrough, not the one to build a weekly cadence around.

None of that requires 2027 data to act on. It's the same conclusion the 2024 report supported, confirmed rather than overturned by everything tracked since.

Personal profiles vs. company pages, for a US mid-market company

LinkedIn puts its own worldwide membership above 1.3 billion, a scale most other B2B channels can't approach. Pew Research, surveying US adults between February and June 2025, found that 25% report using LinkedIn, a share concentrated among college-educated, higher-income professionals, which is precisely the buyer profile most B2B sellers are trying to reach.

The distribution mechanics inside that population favor the individual over the institution. A post published from a personal profile reaches a materially wider audience than the same content published from a company page, and the gap has been widening rather than narrowing as the platform's own recommendation systems lean harder on personal credibility signals. That gap predates 360Brew. The semantic ranker adds a new layer on top of it: it checks whether the person posting has the track record to back up the topic, which raises the cost of ghost-writing a generic executive account and lowers the cost of a founder who actually writes their own material.

For a company in the 10-to-250-employee range, the operational conclusion is the same one the data has supported for several years running: the company page functions as a directory listing, not a growth channel. The founder, the head of sales, and two or three subject-matter specialists posting under their own names will outperform a company account publishing the identical content, every time the comparison has been run.

A four-check pass before you publish

This is the practical version of what the data above actually asks of you. It draws on the same reasoning behind All In's A.U.T.H. framework for reading the LinkedIn feed, reduced to four checks you can run before hitting publish, not a full audit.

Check your topic against your last ten posts. If today's post covers something outside your last three months of publishing, expect a narrower initial test audience. That's the expertise-consistency signal working exactly as designed, and it's not worth routing around.

Pick the format for the job, not out of habit. A checklist, a framework, or a multi-step process belongs in a carousel. A reaction to something that happened this week belongs in text. Don't force a three-paragraph opinion into slide format because carousels perform better on average; a bad carousel underperforms a good text post.

Plan to be present for the first hour. Distribution decisions get made early. If you publish and disappear until evening, you've left the algorithm's first read of your post to whoever happens to comment, which is not a strategy.

Track saves and long comments, not likes. A save is a reader telling the system "I want this again later." A comment over roughly fifty words is a reader spending real effort. Both outrank a hundred one-word reactions in what 360Brew appears to weigh, and both are far better predictors of an actual sales conversation starting in your inbox.

Five mistakes still costing SMEs reach in 2026

Publishing only from the company page. Given the reach gap between personal profiles and company accounts, this is the single most common unforced error among small B2B teams.

Running a post through AI and publishing the first draft. Content that reads templated, with no specific number, name, or dated example anywhere in it, is exactly what a semantic model is built to deprioritize, whether a human or a machine wrote it. That's a different problem from the one LinkedIn's own "Seems like AI slop" reporting button claims to solve, and worth reading separately if you publish with AI assistance.

Stacking hashtags out of old habit. A semantic ranker reads the actual sentence. Three targeted hashtags do no harm; ten do nothing useful and read as dated.

Dropping an external link in the body of the post. The roughly 60% reach penalty documented above is consistent enough across trackers to treat as a fact. Put the link in the first comment, or better, turn the content into a native carousel.

Going quiet after publishing. The engagement window that decides a post's early distribution closes fast. If your calendar has you publishing and then heading straight into back-to-back meetings, you're forfeiting exactly the window that matters most.

What I don't know

I don't know the exact publication date for the van der Blom edition covering full 2026 data. Nobody outside his team does, and any article that states one as fact is guessing.

I don't know what data year his currently sold "Algorithm Report 2026" product actually covers, because it's paywalled and I haven't purchased it. I'm naming that gap rather than papering over it with a number I can't stand behind.

I don't know whether 360Brew's expertise-consistency weighting will hold, strengthen, or get walked back once a full year of data is available to measure it against. Everything in this piece describing 360Brew's effects is drawn from interim, dated trackers through mid-2026, not from an annual audit. That's the honest ceiling of what's knowable right now.

I don't know how the format hierarchy (carousel over text over video) will look once a full year of 2026 posting behavior is measured rather than sampled. The direction has held steady for two years running, which is a reasonable basis for a plan, not a guarantee.

When the next edition ships, whenever that turns out to be, this piece will get an update with the actual numbers attached, and the update will say plainly what changed from the projections above and what didn't. If you'd rather not check back manually, the All In newsletter covers exactly this kind of update the week it happens.

FAQ

Has the van der Blom report covering 2026 LinkedIn data been published?

No, not as of this writing. Richard van der Blom's site currently sells an "Algorithm Report 2026" product without a listed 2027 edition or announced date for one. Based on his publishing cadence since 2019, an edition covering a full year under 360Brew would be expected in the opening months of 2027, but no date has been confirmed.

What was the last confirmed LinkedIn reach and engagement data?

The most recent fully documented dataset covers 2024, published in van der Blom's 2025 report across 1.8 million posts: organic reach down 50% year over year, engagement down 25%, follower growth down 59%. Interim trackers through mid-2026 (Dataslayer, PostUnreel) show the same order of magnitude continuing, without a full annual audit yet available to confirm the exact 2026 figures.

What is 360Brew and how does it rank content differently?

360Brew is LinkedIn's roughly 150-billion-parameter ranking model, which reads posts semantically rather than tallying likes and comments as independent counts. It cross-checks a post's topic against the author's profile and recent publishing history, a mechanism analysts call the expertise mismatch penalty, and distributes more widely to accounts with consistent topical history.

Which LinkedIn format performs best under the current algorithm?

Carousels (native document posts) lead on engagement, tracked at roughly 6.6% to 7% through 2026 against about 2% for text-only posts, largely because they generate two to three minutes of dwell time compared to fifteen to thirty seconds for a typical text post. Video performs respectably but continues to trail carousels on engagement.

Do personal LinkedIn profiles still outperform company pages in 2026?

Yes. Personal profiles consistently reach a wider audience than company pages publishing identical content, and 360Brew's emphasis on individual credibility and posting history has widened that gap rather than closed it. For a small or mid-size company, a founder or executive posting under their own name remains the stronger channel.

Should I trust a specific LinkedIn 2027 statistic I saw online?

Only if it links to a dated, named source. A large share of "2027 LinkedIn algorithm" content circulating online extrapolates from 2024 or 2025 data without saying so, or restates unverified figures with no attached study. If an article can't point to who published a number and when, treat the number as a guess.

Sources

  1. Richard van der Blom, Algorithm Insights Report, Just Connecting, 2025 edition (2024 data, 1.8 million posts), richardvanderblom.com.
  2. Richard van der Blom, richardvanderblom.com, accessed August 2026: current product listing ("Algorithm Report 2026," corporate licensing), no 2027 edition announced.
  3. Dataslayer, LinkedIn Algorithm 2026: What's Working Now, published February 12, 2026, updated July 2026, dataslayer.ai.
  4. PostUnreel, LinkedIn Carousel Engagement 2026 Data, 2026, postunreel.com.
  5. UpGrowth, LinkedIn Algorithm 2026 Explained: What 360Brew Means for Reach and Growth, March 2, 2026, upgrowth.in.
  6. Project Neon, LinkedIn Algorithm 2026: How 360Brew Is Redefining What Works, January 2026, project-neon.com.
  7. Thomas Pettauer, LinkedIn 360Brew and the New Physics of Visibility, January 26, 2026, pettauer.net.
  8. Pew Research Center, Social Media Fact Sheet, survey conducted February to June 2025, published June 18, 2025, pewresearch.org.

All In: read the algorithm data before you build a calendar on top of it

Most "2027 LinkedIn predictions" you'll read this year cite nothing dated and nothing named. This one tells you exactly what's confirmed, what's tracking mid-cycle, and what we're still waiting on.

All In is the B2B media that decodes LinkedIn, expert blog, weekly podcast and newsletter for SME leaders and sales directors who want to turn LinkedIn into measurable growth. An original creation by Patrick de Carvalho, on LinkedIn since 2004. Motto: "I Never Lose."

Discover All In and get what matters on LinkedIn, every week.