A useful way to think about LinkedIn today is as a marketplace.
A few years ago, it felt relatively calm. People posted occasionally, shared career updates, and wrote the odd thoughtful piece. You could scroll your feed and get a sense of who people actually were.
Now, it feels very different.
The marketplace is louder, more active, and much more crowded. And increasingly, many of the stalls look… similar.
That shift is largely because of AI. Writing tools have made it incredibly easy to produce content quickly and at scale. Some estimates suggest that over half of long-form posts on LinkedIn may now be AI-generated (Gillham, 2024). Even viral posts are increasingly suspected to be AI-written (Knibbs, 2024).
So the question isn’t just what to post anymore. It’s how the algorithm decides what’s worth showing.
How the algorithm decides what gets seen
If LinkedIn is a marketplace, the algorithm is what decides which stalls get foot traffic.
It doesn’t distribute your post widely right away. Instead, it starts by showing it to a small portion of your network. Based on how that group reacts, the post either gets pushed further or fades out (Newberry & Christison, 2025).
But not all engagement is equal.
Likes still matter, but comments carry more weight, especially when they lead to conversations (Newberry & Christison, 2025). A post with thoughtful replies and back-and-forth discussion signals something deeper than passive approval.
Then there’s dwell time, which has become one of the most important signals. LinkedIn has explicitly shared that how long users spend engaging with a post influences its ranking (Zhang et al., 2024).
In simple terms, the algorithm is asking: Did this post make someone stop and actually pay attention?
What AI has changed
AI hasn’t just increased the volume of content; it has standardised it. If you scroll through LinkedIn today, you’ll notice patterns: similar hooks, short, spaced-out sentences, and emotional storytelling arcs. These formats work, which is why AI tools replicate them so easily.
The result is a feed full of content that is technically “good,” but often hard to distinguish. At the same time, posting frequency has gone up because AI reduces the effort required to create content. More supply means more competition, which makes the algorithm more selective, not less.
Even virality has become less meaningful. A post can perform well because it’s structurally optimised, not necessarily because it reflects a real experience (Knibbs, 2024). So the algorithm has to look beyond surface-level quality.
A subtle shift toward authenticity
LinkedIn’s response to this shift has been gradual but telling. For instance, the platform has introduced labels for AI-generated content, signalling a move toward transparency (PS Digitise LLP, 2024).
More importantly, the way content performs suggests that authenticity is becoming a stronger signal. Posts that are grounded in specific experiences or clear points of view tend to generate more meaningful engagement, and therefore more reach.
This aligns with how the algorithm works at a deeper level. It prioritises relevance to your network and niche, meaning generic, one-size-fits-all content is less likely to perform well (Newberry & Christison, 2025).
It also filters out low-quality or spammy posts more aggressively, which includes repetitive or engagement-bait formats (Newberry & Christison, 2025).
So while AI can replicate structure, it still struggles with context and nuance, and that’s increasingly what stands out.
What actually works now
In this environment, the content that performs well isn’t necessarily the most polished. It’s the content that holds attention and invites interaction.
Posts that increase dwell time, such as well-structured text or carousel-style content, tend to perform better because they keep users engaged longer (Zhang et al., 2024).
Content that sparks discussion rather than just collecting likes also travels further. This is why posts with strong opinions, questions, or relatable experiences often outperform generic motivational content.
On the flip side, posts that push users off the platform, such as those with external links, tend to have reduced reach (Newberry & Christison, 2025).
Over time, consistency matters too. Creators who regularly engage with their network and build credibility tend to see compounding results, because the algorithm learns that their content generates meaningful interactions.
Where is this heading
If you step back, the bigger shift is this: the algorithm is moving from ranking content based on quality alone to ranking it based on signals of genuine human engagement.
AI has made it easy to create content. But it hasn’t made it easy to create a connection. And that’s where the opportunity lies.
In a marketplace where many stalls look the same, the ones that stand out are usually the ones that feel more specific, more opinionated, and more human.
Because ultimately, the algorithm isn’t just distributing content, it’s learning which voices people trust enough to stop and listen to.
