Is LinkedIn's algorithm rewarding "male-coded" content?

Something unusual has been happening on LinkedIn, and it's begun to raise some questions about what really drives reach on the world's largest professional platform.

In recent months, some users have been running quiet, side-by-side experiments after noticing a drop in visibility and reach.

Some women decided to change the listed gender on their profiles and adjust their tone to sound more "male-coded," while men began to test the opposite approach.

The results have since flooded the feed with posts that reveal dramatic spikes in impressions, views and engagement, sparking debate about algorithmic bias and its potential impact on the 23-year-old platform's 1.3 billion users.

Algorithms

Now, it's no secret that algorithms operate in ways that are nearly impossible to understand or predict.

Especially as machine-learning models grow more complex, relying on millions of data points and opaque decision-making processes.

But when you layer these experimental findings into a platform that AI-assisted writing tools and other performative expectations are already reshaping, the conversation around what the future of digital interaction looks like gets that much more complicated.

At what point does optimising our voice for algorithmic gain shift from a smart strategy to a loss of authenticity? And what, if anything, can we do instead?

From networks to signals

In a recent blog post, LinkedIn stated that its algorithm and AI systems do not factor demographic attributes, such as age, race or gender, into decisions about the visibility of profiles or content in a user's feed.

It's a point also echoed by LinkedIn strategist Sanjiv Ramjee. He explains:

"That message is baked into everything that LinkedIn does. I've seen that from the inside in terms of how they develop their products, how they mitigate bias within their products, and so on."

Surprised to see the reports of users experiencing more favourable outcomes after changing their gender from female to male, Sanjiv says that for what he knows about the latest changes to the algorithm, there could have been several reasons for these kinds of results.

"What we're seeing now is that the algorithm is now going to look at things such as your LinkedIn profile."

Instead of looking for quantitative metrics like followers or connections, it now critically assesses skills, experience, profiles and posting behaviour to determine relevance.

"It's then going to use a lot of that information to try and get your content in front of people who are similar to you, or people who are going to resonate with that content."

He adds:

"If LinkedIn detects a change on your profile, then naturally it wants to highlight what that change is.

"So if you're posting something at a time when you're changing something on your profile, that could be a reason why your posts are getting more traction at a particular point in time."

However, as more of these experiments – and their findings among men, women and marginalised communities – become increasingly amplified across the platform, the discussion has moved beyond anecdote.

A petition has begun to circulate calling for LinkedIn to address the issue directly, and some communities, like the European Women's Management Development Network, have also started to organise online events with hundreds of participants to document and discuss the impact.

Proxy bias

Looking into the matter further, independent analysis from ethical AI expert Martyn Redstone also suggests these patterns may stem from proxy bias.

This is where algorithms trained on historical data inadvertently favour certain language, topics and career signals that correlate with gender.

As Sam Robson, founder of the better web co., explains, mainstream platform algorithms are rarely programmed to discriminate.

However, because they rely on machine learning, "if the data they train themselves on shows bias, that can reflect in their output."

For social media platforms, that training data is largely based on user behaviour, meaning that existing societal bias – whether related to gender, race or sexuality – can become embedded in algorithmic systems over time.

Sam says:

"If someone on Facebook keeps clicking sexist posts, the algorithm will show them more of that kind of content.

"If a recruiter keeps approving CVs from men and rejecting those from women, there's a risk the algorithm learns to only suggest CVs from men.

"The algorithms don't innately understand right and wrong, so if unchecked they'll pick up racist, sexist and homophobic tendencies, because unfortunately these biases remain present in society."

This, Sam argues, is when platforms must intervene.

He references a familiar example with Google autocomplete – a feature that suggests words as you type based on the popularity of searches – which Google is forced to constantly monitor and filter. "They don't always get it right."

The cost of being unseen

Perhaps these interpretations point to a more uncomfortable reality – one where visibility isn't only increasingly shaped by what you say online, but by how legible your professional identity is to the system interpreting it.

But as users continue to navigate this narrow line between optimising for reach and staying true to how they actually work, speak and lead, does the question become less about performance and more about sustainability?

And if visibility is essential for growth, is it worth changing how you present yourself to "fit" what the platform seems to reward?

For Sanjiv, the answer lies in shifting focus away from short-term metrics and toward long-term relevance.

"Think less about the vanity metrics and instead think about your ideal profile as a business.

"Who are you trying to reach? Who are you trying to add value to and continually produce content that speaks to their wants and needs?"

Indeed, curating your content to line up with what you think the algorithm might favour is a tricky game, and one that could feel less like a strategy and more like a matter of luck.

Sanjiv says:

"If you're producing content for the algorithm so it maximises your reach and the people who are going to see your content, what's to say that that changes in the next couple of months?

"And now your strategy has to radically change, because you're not reaching now the same other people that you were before?"

Creating LinkedIn content that genuinely resonates

Instead, Sanjiv lays out several key points to consider when building an online presence and creating content that genuinely resonates with your intended audience.

1. Treat your LinkedIn profile as something dynamic rather than static. "You should continually look to update your profile, whether that be an up-to-date picture, up-to-date customer wins, up-to-date skills or events that you're holding."

2. Stay true to yourself and your audience. "If you're trying to be somebody else, you're trying to be too polished and too corporate," he adds, suggesting that more and more users are beginning to detect and deter from that kind of content.

3. Experiment with AI, but know your limit. There are a number of AI tools that can support with generating and clarifying ideas, which can be a game-changer for many, Sanjiv explains.

   But "if you're writing a post, don't get ChatGPT to write your post for you and then copy and paste – that's the differentiation between sort of using AI and not using AI."

Read more in this series

• Hiring for a future shaped by AI and change

• The mindset that makes Britain's innovation possible

Make your small business more productive with the free Tech Hub tool

Get tailored recommendations, join virtual workshops, connect with expert advisers and find practical resources. Take me to Tech Hub now