What happened
LinkedIn strategist Melanie Goodman published a detailed 2026 algorithm breakdown this month showing that first-degree connections now account for roughly 31% of what the average user sees in their feed. The rest is served by an interest graph that surfaces posts based on inferred topics, engagement history, and what LinkedIn calls a depth score. Generic content that could have been written by anyone gets suppressed; specific, author-attributed observations get carried past the network boundary. Untagged posts also now outperform posts with hashtags by 5 to 10%, a full inversion of pre-2024 behavior. The core ranker itself has not been rewritten between April and August, but the compounding effect of the interest-graph shift, the AI-slop suppression pass, and the external-link penalty is finally showing up in creator reach dashboards.
Why the LinkedIn feed shift matters now
MagnetizeX builds founder visibility systems for B2B firms.
For founders, the working assumption for a decade was that connection count was the leading indicator of reach. It no longer is. A post from a founder with 2,400 connections and tight topical authority now outperforms a 40,000-connection founder posting off-topic thought-of-the-day content. That flips how you plan a Q4 content push: you optimize for a narrow subject the algorithm can classify you against, not for total network size. The interest graph does not care who follows you. It cares whether the post could have been written by anyone else on the platform.
- Audit your last 20 posts for topical drift
Paste them into a spreadsheet and tag each one with the primary subject. If more than three distinct topics appear, the interest graph cannot classify you consistently. Cut to one or two subjects for the next 30 days and re-measure impression-per-post before assuming reach loss is a platform problem.
- Stop chasing connection count as a KPI
Replace it with two metrics the interest graph actually cares about: median impressions per post over the last 30 days, and the ratio of second-and-third-degree views to first-degree views. If out-of-network views are climbing while impressions stay flat, the graph is working. If both are flat, the topic is drifting.
- Remove hashtags from your next five posts
Publish the same content shape you normally would, with zero hashtags. Compare reach against your last five hashtag-tagged posts of similar length and topic. Goodman's data and LinkedIn's own product changes both point the same direction: hashtags are now a mild negative signal, not a distribution amplifier.
What to do this week
Open Shield Analytics or Taplio and pull your last 30 days of post-level data. Sort by impressions, isolate the top three, and identify the single subject they share. That is your working topical anchor for September. Post four times against that anchor next week, hashtag-free, and treat any post outside it as an experiment, not a staple.