LinkedIn Was Never Built to Help You.
It Was Built to Match You.
I used to think the "LinkedIn is just showing me my competitors" complaint was a bit of a cop-out. I'd heard it from friends in a handful of industries (consultants, insurance people, one colleague who wouldn't stop bringing it up at ) and my first reaction was always some version of “use the platform better.” Optimize your headline. Post more. Engage with the right content. The tools are there; you're just not using them right.
Then I started paying attention to my own feed as a media production professional, and I stopped being able to make that argument with a straight face.
The design problem, not the user problem
Here's the claim, stated plainly:
LinkedIn's connection-suggestion engine was never built to connect you with the people who actually grow your business.
It was built to connect you with people who look like you on paper, same job title, same industry tag, same geography, because that's the signal the underlying model is tuned to weight most heavily. In the recommendation-systems literature, this is called homophily: similar people cluster with higher probability than dissimilar people, and "People You May Know"-style engines are built directly on that principle.
Take a financial advisor. Their entire business depends on meeting people who need financial planning, people two or three tax brackets removed from the advisor world, in industries the advisor doesn't work in, at firms the advisor has never heard of. But log into LinkedIn as a financial advisor and the platform's default instinct is to hand you a stream of other financial advisors. Same for insurance agents. The suggestion engine isn't malfunctioning here; it's doing exactly what it was tuned to do. It's just optimizing for the wrong relationship. It treats "professional network" as "people with your job title," when for an advisor, the only network that matters commercially is the one made of people who are not in that job.
And I don't think this is a blind spot LinkedIn stumbled into. In September 2022, LinkedIn researchers, working with teams from Stanford, MIT, and Harvard, published a study in Science documenting experiments the company ran on more than 20 million users over five years, deliberately adjusting the mix of "weak tie" and "strong tie" suggestions in the People You May Know algorithm and measuring the effect on job mobility. That study exists because LinkedIn can tune this system with precision, toward a chosen outcome, at massive scale. Which means the peer-matching default the rest of us live with isn't a technical limitation. It's a choice they've demonstrated they know how to make differently.
The accountability problem
The second part of this is worse, because it's not about design intent. It's about whether the system answers to the person using it.
Every "People You May Know" card has an X in the corner. Click it and LinkedIn tells you, implicitly, that it has registered your feedback and will adjust. In my own experience, and in the experience of the friends I mentioned earlier, that adjustment doesn't reliably happen. The same category of person, the same occupation, the same look-alike profile, keeps resurfacing. There's no published research confirming whether or how much weight a dismissal carries in the backend model, because LinkedIn doesn't publish that. What I can say is that the interface asks for a signal and gives no visible evidence that signal changes anything, which is a specific kind of design failure: it's not that the platform ignores you, it's that it built you a button whose only confirmed function is to make you feel like you did something.
The "People Also Viewed" setting makes the opacity even more visible. LinkedIn's own Help documentation still describes it as an adjustable toggle. Independent reporting says LinkedIn quietly removed that toggle in early 2024, with no announcement, and that the module is now permanent on every profile. I'm not going to resolve that contradiction for you, because I can't, and that's the point. When a platform's own documentation can't agree with outside observers on whether a control exists, "user control" was never really the operating model. Engagement was.
Where I've seen this exact failure before
A few weeks ago I went looking for a way to stop LinkedIn from feeding me other photographers and media producers, using Google's AI Mode. What I got was a long, increasingly specific set of instructions: click here, then here, toggle this setting, change your industry classification, delivered with total confidence. Several of those settings didn't exist where the AI said they did. When I corrected it, it didn't say "I'm not sure of LinkedIn's current layout." It apologized and produced an equally specific, equally wrong set of steps. When I told it the dismiss button wasn't working, it agreed the button was likely theater and told me to install an ad blocker instead of fixing anything.
That's the same failure as the X button, wearing a different interface. A chatbot hallucinating a settings menu and an algorithm ignoring your dismissal are both systems simulating responsiveness without being answerable for it. You ask, you get an answer that has the shape of help (confident, specific, delivered instantly) and nothing about your actual situation changes. AI search tools and social platforms increasingly share this instinct, whether or not anyone designed it on purpose: tell the person what sounds like an answer, because a satisfied-looking interaction is what keeps them coming back, whether or not it was accurate or effective.
What LinkedIn actually says it's for
LinkedIn's own mission statement is to connect the world's professionals to make them more productive and successful. That's a fine mission. It is not, in practice, what the "People You May Know" engine is optimized to do for anyone whose business depends on meeting people outside their own occupation. The gap between that stated mission and the mechanism sitting underneath it isn't an accident of scale. It's the product working as designed, just not for the person using it.
Sources
The following sources were located using AI-assisted web search and are provided so readers and researchers can independently verify and trace the claims made above. Their inclusion here does not constitute an endorsement of any source's overall reliability; readers should evaluate each source on its own merits.
Rajkumar, K., Saint-Jacques, G., et al. "A causal test of the strength of weak ties." Science, September 15, 2022. https://www.science.org/doi/10.1126/science.abl4476
MIT News. "The power of weak ties in gaining new employment." September 15, 2022. https://news.mit.edu/2022/weak-ties-linkedin-employment-0915
Stanford Report. "The real strength of weak ties." https://news.stanford.edu/stories/2022/09/real-strength-weak-ties
UPI. "LinkedIn ran secret experiments on 20M users in study on the strength of social ties." September 25, 2022. https://www.upi.com/Science_News/2022/09/25/linkedin-ran-secret-experiments-20-million-users-strength-weak-social-ties/7411664131874/
McPherson, M., Smith-Lovin, L., Cook, J.M. "Birds of a Feather: Homophily in Social Networks." Referenced via: "Semantic homophily in online communication: evidence from Twitter." arXiv. https://arxiv.org/pdf/1606.08207
"A Survey of Link Recommendation for Social Networks." arXiv. https://arxiv.org/pdf/1511.01868
About LinkedIn (official mission statement). https://about.linkedin.com/
LinkedIn Help. "Remove or Add the People Also Viewed Box." https://www.linkedin.com/help/linkedin/answer/a543958/
DemandBird. "LinkedIn 'People Also Viewed': The Toggle Was Removed in 2024." https://demandbird.com/resources/linkedin-people-also-viewed/
LinkedIn News. "Jobs on the Rise 2026: The 25 fastest-growing roles in the U.S." https://www.linkedin.com/pulse/linkedin-jobs-rise-2026-25-fastest-growing-roles-us-linkedin-news-dlb1c
Forbes. "Future-Proof Your Career With LinkedIn's 2026 Fastest-Growing Jobs List." January 14, 2026. https://www.forbes.com/sites/juliakorn/2026/01/14/future-proof-your-career-with-linkedins-2026-fastest-growing-jobs-list/
Note: The AI Mode conversation referenced in this piece is the author's own, paraphrased from a personal transcript, and is not independently linked here.