Every mechanic in the game, with how much evidence stands behind it. Proven means LinkedIn published it. Measured means an independent study with a published sample. Disputed means credible studies disagree. Invented means it circulates widely and traces to nothing. Where the mechanism is sourced but the size of the effect is ours, the entry says so.
Play the gameOne researcher, describing his own posts, in one interview. It became a statistic. So did this title.
van der Blom, Creator Science interview (at 23:02): an estimate that about 80% of the early comments on his own posts are AI-written. An anecdote about one account, not a study finding. It circulates as a statistic. Read the source
LinkedIn says it is filtering engagement bait and quotes "Comment 'Yes' if you agree" as an example (Jurka, March 2026). The 0.6x is ours.
Jurka, Updates to The LinkedIn Feed Focusing on Authentic, Relevant Conversations, 12 Mar 2026 Read the source
Your reach was cut this week by something earlier: a pod, or engagement bait twice in three weeks. LinkedIn says it acts on both (Jurka, March 2026). The size of the cut is ours.
Jurka, Updates to The LinkedIn Feed Focusing on Authentic, Relevant Conversations, 12 Mar 2026 Read the source
LinkedIn says it is making engagement pods ineffective (Jurka, March 2026). The comments were real people being polite. Your next two weeks of reach are halved; that size is ours.
Jurka, Updates to The LinkedIn Feed Focusing on Authentic, Relevant Conversations, 12 Mar 2026 Read the source
LinkedIn says it is limiting what automated comments and pods can do for a post (Jurka, March 2026). Forty comments, no distribution.
Jurka, Updates to The LinkedIn Feed Focusing on Authentic, Relevant Conversations, 12 Mar 2026 Read the source
LinkedIn's retrieval matches posts on meaning, and its language model relates topics nobody told it were related (Danchev, March 2026). The 8% is ours.
Danchev, Engineering the next generation of LinkedIn's Feed, LinkedIn Engineering, 12 Mar 2026 Read the source
Your headline, company and industry are part of the text that describes every post you write (Danchev, March 2026). The size of the effect is ours.
Danchev, Engineering the next generation of LinkedIn's Feed, LinkedIn Engineering, 12 Mar 2026 Read the source
Text posts reach 1.07x a typical post and draw 0.78x the engagement (AuthoredUp, 3M posts, each profile against its own median).
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Images reach 1.20x a typical post and draw 1.33x the engagement, the best engagement of any format (AuthoredUp, 3M posts).
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Documents reach 1.39x a typical post, second only to polls, and draw 1.30x the engagement (AuthoredUp, 3M posts).
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Video reach fell 36% in a year and video now reaches 0.86x a typical post; clips under 30 seconds do a little worse (AuthoredUp).
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Videos over three minutes reach 1.21x the average video (AuthoredUp, 37k videos). Long-form still works. It also costs four hours.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Polls reach 1.78x a typical post and draw 0.37x the engagement (AuthoredUp, 3M posts). Wide and shallow, and pipeline follows attention, not reach.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Articles reach 0.69x a typical post and draw 0.44x the engagement (AuthoredUp). Three hours for the least-read format apart from reshares.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Reshares reach 0.29x a typical post and draw 0.22x the engagement (AuthoredUp). Cheap, and priced accordingly.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
Posts that opened with the point did about 10% better than posts that opened with a scene, in my own 107 posts. A small effect, from one account.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
Opening with a scene instead of the point cost about 10% in my own 107 posts. One account, and engagement rather than reach.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
Named people, companies and numbers drew about 1.2x the engagement of generic advice in my 107 posts. The gap has narrowed in 2025 and 2026.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
Personal stories with real stakes were 12% of my posts and 53% of my engagement. The game counts them at 2x, discounted for the congratulations that inflate the raw figure.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
Promotional posts drew about 0.75x the engagement of my median post, across 52 of them.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
A specific, answerable closing question drew a median of 9 comments against 2 without one, in my posts. Eight posts, so treat it as direction.
Jason Nellis, 107 of his own LinkedIn posts, 2014 to 2026. One account, engagement not reach, single coder.
Measured, and a trap. Polls reach 1.78x a typical post and draw 0.37x the engagement (AuthoredUp, 3M posts). Everyone saw it. Nobody wrote.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
AuthoredUp found hashtags do not help, and that more than six "can seriously sabotage your reach". They published a direction, not a number; the 15% here is ours.
AuthoredUp, Best Performing Content on LinkedIn: 3M+ posts, personal profiles, Mar 2025 to Feb 2026. Vendor data, relative to each profile median. Read the source
This one works. Interact reports that 40.1% of quiz starts become leads, across 100M+ leads. Interact sells quizzes. You are playing one.
Interact, Quiz Conversion Rate Report (updated 8 Sep 2026). Vendor platform data, 100M+ leads. Read the source
Van der Blom says reach for active creators is down about 60% over two years, speaking about his paid 2026 report. LinkedIn says it chose relevance over reach.
van der Blom, Algorithm Insights 2026 (paid report). Figures from his Creator Science interview and his public LinkedIn newsletter, 26 Jul 2026. Read the source
Disputed. Van der Blom measures 16% less reach for posts with an external link; Ordinal says personal profiles see almost none. Modelled here as readers leaving for the link.
van der Blom, Algorithm Insights 2026 (paid report). Figures from his Creator Science interview and his public LinkedIn newsletter, 26 Jul 2026. Read the source
Disputed. Nobody has published how a link in the first comment behaves. Ordinal, which sells a first-comment tool, says personal profiles barely see a link penalty either way.
Ordinal, LinkedIn link penalty study, 900k+ posts. Vendor with a commercial interest; method unpublished; says personal profiles see almost no penalty. Read the source
The +21% traces to nothing anyone has published; a 2026 review of LinkedIn claims could not find a source for it. It cost you half an hour.
Dhélin, The LinkedIn Algorithm 2026: what is proven, what is measured, what is invented, Fast Growth Advisors, Jul 2026 Read the source
The 60% link penalty comes from vendor blogs, not data. The link effect itself is disputed: van der Blom measures 16% less reach, Ordinal almost none for personal profiles.
Dataslayer, LinkedIn algorithm February 2026 (example of the folklore: 60% link penalty, 0.07% polls) Read the source
No study measures it, and LinkedIn's own examples of engagement bait do not mention it. It did nothing. Nobody replied to the question either.
No primary source or study found (checked Sep 2026). LinkedIn's own engagement-bait examples (Jurka, 12 Mar 2026) do not include it. Read the source