Richard's unfair advantage is that he owns the dataset
Everyone has opinions about the LinkedIn algorithm. He has 1.3 million posts and 1,400 hours of research.
Richard van der Blom publishes the Algorithm Insights Report, a 200-page annual study now in its seventh edition, built on 1.3 million posts and more than 1,400 hours of research. It is the document most of the LinkedIn industry quotes when it makes a claim about reach. Across the 97 posts we analyzed, published between April and July 2026, he posted 6.2 times a week and averaged 273 reactions. 56 of those posts mention the algorithm directly.
The account is a distribution layer for a dataset nobody else has. Proprietary-data authority is when your credibility rests on a measurement only you have run, so every claim you make is checkable and nobody can contradict you without doing the work themselves. He does not argue about the algorithm. He reports what 1.3 million posts did.
'Here's what I think the algorithm rewards.' Plausible, unfalsifiable, forgotten by Friday.
'In 2024, 70% of distribution went to your connections. In 2026, just 10%.' Checkable.
“Harsh? Maybe. But the data doesn't lie.”
— From his most-reacted post, on connections no longer driving distribution (1,458 reactions)
Five findings that repeated across 97 posts
- The reach is low and the discussion is high. A median of 242 reactions on 272,000 followers, with a 42.1% comment ratio, roughly seven times the norm.
- Interest is his second reaction at 14.4%, the highest in this series. People are here to learn something specific.
- Nobody forwards him. A mean of 7 reposts, one of the lowest we have measured, despite research that would seem eminently shareable.
- He works six and a half days. 89 of 97 posts land Sunday to Friday, with Saturday the only lighter day at 8.
- The research is the spine. 56 posts mention the algorithm and 38 reference the report itself.
The numbers behind the account
This is the most striking gap between authority and reach anywhere in this series, and his own research explains it.
Richard's median post earns 242 reactions and his mean is 273, on an audience of 272,000. That is the lowest reach-per-follower of any account we have taken apart. Only one post in 97 cleared 1,000, 75 landed between 100 and 500, and 10 finished under 100. For the person the industry treats as the authority on LinkedIn distribution, that is a genuinely uncomfortable set of numbers.
That shift is the single most important thing in this teardown for anyone reading it. We cover the mechanics in our guide to how the LinkedIn algorithm works.
How the reach is distributed
When he posts
Where the engagement comes from
The top posts
| # | Post | Reactions | Comments | Reposts |
|---|---|---|---|---|
| 1 | Your connections don't matter anymore | 1,458 | 471 | 81 |
| 2 | Not all LinkedIn posts die the same way | 755 | 145 | 31 |
| 3 | LinkedIn: We're fighting AI slop | 711 | 358 | 6 |
| 4 | I called my dad yesterday for Father's Day | 664 | 172 | 1 |
| 5 | 5 trends, based on 12,000+ creators since March | 645 | 276 | 19 |
| 6 | The Algorithm Insights Report 2026 is LIVE | 603 | 216 | 39 |
Curious how your own reach compares to your follower count? Run your profile through our free LinkedIn analyzer.
The six content pillars
Four of them come out of the dataset. The other two are what stop it reading like a spreadsheet.
One number from the research that overturns something the industry believes.
A pattern named and split into types, so readers can classify their own posts.
A scripted dialogue with LinkedIn itself, usually about something that happened to him.
Rare, entirely unrelated to the research, and among his highest reach.
What has changed since the last report, with the sample size stated up front.
The annual publication, sold on scale: pages, hours, editions, posts analysed.
Pillar 1: The headline finding (the reach engine)
Why it works: His biggest post by a factor of two, and 471 comments. The three-year series does the persuading: 70%, 54%, 10% is a trend nobody can argue with, and 'Read that again' is the only editorialising in the whole post. The data is confrontational so he does not have to be.
Pillar 2: The taxonomy (the most useful)
Why it works: Naming the patterns is what makes research usable. 'The Rocket, 45% of posts' gives the reader a label for something they have observed but never had a word for, and a percentage tells them how common their own experience is.
Pillar 3: The platform satire (the comment engine)
Why it works: 358 comments on 711 reactions, a 50% ratio. The dialogue format lets him complain about the platform without sounding aggrieved, and 'That's an infographic. With bar charts.' is the line everyone quoted back. Note the 6 reposts: people replied rather than forwarded.
Pillar 4: The personal post (the deepest)
Why it works: Fourth-biggest post on an account otherwise made entirely of data, and one repost. He states the hardest facts in fragments, four years, two lost pregnancies, medical complications, without elaborating on any of them. Restraint is what stops a post like this reading as a bid for sympathy.
Pillar 5: The trend update (the freshness)
Why it works: He leads the first trend with his own losses, down 22% on his own content. Publishing a finding that indicts your own performance is the strongest credibility signal available to a researcher, and it costs nothing because the data was going to say it anyway.
Pillar 6: The report launch (the business end)
Why it works: The pitch is entirely scale: pages, chapters, tactics, posts, hours, editions. Six numbers and no adjectives. For a research product that is the correct sell, because the only real question a buyer has is whether enough work went into it.
The hooks that earned the click
Most of them state a finding that contradicts something the reader believes.
'Your connections don't matter anymore. Harsh? Maybe. But the data doesn't lie.'
'Not all LinkedIn posts die the same way.'
'LinkedIn: We're fighting AI slop. / Me: Finally. Let's go.'
'5 trends worth knowing, based on +12,000 creators since March.'
'LinkedIn owes you an explanation. Your follower count went up. Your reach went down.'
'The LinkedIn algorithm no longer is a black box.'
Almost every hook contains either a number or a named pattern, which is what separates research content from commentary. For the mechanics of openers, our guide to writing LinkedIn hooks goes deeper, and you can pressure-test your own first line in the free hook generator.
A voice that lets the numbers be rude
He states findings flatly, names his own losses, and keeps the editorialising to a minimum.
- Leads with the figure, not the interpretation. Three-year series, percentages, sample sizes in the first two lines.
- States the sample every time. 1.3 million posts, 12,000 creators, 40,000 carousels, 80,000 profiles.
- Publishes findings that indict his own performance, including a 22% drop in his own reach.
- Writes at moderate length. A 258-word median, enough for a finding plus its caveat.
- Uses almost no hashtags, 5 across 97 posts.
- Works Sundays. 14 of 97 posts land on a Sunday and only 8 on a Saturday.
The research post structure is the transferable part, and it works for anyone sitting on a measurement their industry lacks.
[The overturned belief, stated flatly: 'Your connections don't matter anymore.'] [The pre-empt: 'Harsh? Maybe. But the data doesn't lie.'] [The series, one line per year, so the trend is undeniable] [The counter-movement: what rose while that fell] [The implication, in one sentence, and then stop]
What he does, and doesn't, do
- Own the dataset, then report it plainly
- State the sample size in every claim
- Publish findings that hurt his own numbers
- Name his patterns so readers can classify themselves
- Sell the report on scale rather than adjectives
- Arguing about the algorithm
- Hashtags, in 92 of 97 posts
- Video, at 2% of output and his weakest format
- Overstating what the data shows
- Making the personal posts about the research
The uncomfortable observation this account invites is that authority and reach have decoupled. He is the most-cited source on LinkedIn distribution and his median post reaches 242 people's reactions. His own research explains why: a follower count built before 2026 is worth a fraction of what it was. CaptureFlow is an AI content agent that turns your expertise into weeks of on-brand content for every platform. You capture one idea in 5 minutes, a finding and a chart, and CaptureFlow, trained on your voice and your past posts, reshapes it into a LinkedIn post, an X thread, a carousel, or a short video, so one piece of research reaches every channel rather than one. See how the AI content agent works.
The system underneath the posts
One annual study, broken into a year of posts, sold as a single product.
The research funnel
- 1The study runs1.3 million posts, 1,400 hours, seventh edition.
- 2Findings are released one at a time56 of 97 posts reference the algorithm.
- 3Each finding is checkableSample size stated, percentages given, no interpretation added.
- 4The industry cites himThe report becomes the reference everyone quotes.
- 5The report sells on that reputationAnd funds the next edition.
This is the same evidence-first model we mapped in the Lily Ray playbook, though she earns 30 reposts a post to his 7, largely because her evidence is screenshots rather than prose. For marketing teams sitting on proprietary data, the pair together show that owning the finding is only half the job.
Your 30-day challenge
Twelve posts, each reporting something you measured rather than something you think.
- Days 1-3: Pick something in your field everyone has opinions about and nobody has measured
- Days 4-5: Measure it, however small the sample, and write the sample size down
- Days 6-7: Publish the headline finding with the number in the first two lines
- Days 8-10: Split your finding into named types with percentages
- Days 11-12: Give each type a label a reader can apply to themselves
- Days 13-14: Publish a finding that is inconvenient for you
- Days 15-17: Turn your best finding into a chart somebody could paste into a deck
- Days 18-19: Compare its reposts against the text version
- Days 20-21: Pre-empt the objection in one line, then decline to argue
- Days 22-24: Publish one post entirely unrelated to your research
- Days 25-27: Do not connect it back to your subject
- Days 28-30: Compare your reach-per-follower against last month, not your raw reach
The stop-doing list
| Stop doing | Do this instead |
|---|---|
| Opinions about your industry | Measurements of it, with the sample stated |
| Describing a pattern | Naming it, and giving it a percentage |
| Hiding findings that hurt you | Leading with them; it is your strongest credibility signal |
| Defending your data | Pre-empting the objection once, then stopping |
| Dense text for a chart-shaped finding | A chart people can forward into a deck |
| Measuring raw reach | Measuring reach relative to your follower count |
The metrics to track weekly
| Metric | Benchmark to aim for |
|---|---|
| Interest reaction share | 10%+ (his is 14.4%, the highest here) |
| Comment-to-reaction ratio | 20%+ organic (he sits at 42.1%) |
| Reposts per post | His 7 is the gap; aim far higher |
| Posts stating a sample size | Every research post |
| Reach per 1,000 followers | The number that reveals the real problem |
| Posts per week | 6, Sundays included |
The takeaways
- 01Own the measurement. Richard van der Blom's authority rests on 1.3 million posts and 1,400 hours of research nobody else has run.
- 02State the sample every time. 1.3M posts, 12,000 creators, 40,000 carousels. The number is the credibility.
- 03Lead with the finding that hurts you. He opened a trends post with a 22% drop in his own reach.
- 04Name your patterns. 'The Rocket, 45% of posts' gives readers a label for something they had noticed but could not describe.
- 05Pre-empt the objection once, then stop. 'Harsh? Maybe. But the data doesn't lie.' is the whole defence.
- 06Watch reach per follower, not reach. His own research shows connection-based distribution fell from 70% to 10% in two years.
Frequently asked questions
- Who is Richard van der Blom?
- He publishes the LinkedIn Algorithm Insights Report, a 200-page annual study now in its seventh edition, built on 1.3 million posts and over 1,400 hours of research. It is the document most of the industry cites when making claims about LinkedIn reach.
- What kind of post performs best for Richard van der Blom?
- Findings that overturn a common belief. His top post, reporting that connection-based distribution fell from 70% in 2024 to 10% in 2026, earned 1,458 reactions and 471 comments, roughly double anything else in the sample.
- Why is his reach low for 272K followers?
- His own research answers it. Connection-based distribution dropped from 70% to 54% to 10% across three years while the Interest Graph rose to 50%, so a follower count built before 2026 is worth a fraction of what it was. His median post earns 242 reactions.
- How do you turn one research project into a year of posts?
- Capture findings continuously as the analysis runs rather than mining a finished report. CaptureFlow turns one 5-minute capture into a week of native posts, carousels and infographics across platforms.
