LinkedIn Likes Me Better as a Man

A look at gender code switching and the LinkedIn algorithm.

I’ve had a wild week. Before I get into the details, I want to be clear about something: I am not trying to become the gender-bias person on the internet. That’s not the work I’m here to do, because my passion is mental health. That’s where I spend my time, and that’s the work I care about. The only reason I’m talking about gender at all is because it started getting in the way of doing my actual job.

(And isn’t that the insidious nature of gender bias? That you can’t talk about it without being the gender-bias person, but it keeps you from doing the work you actually care about.)

I built my business on LinkedIn. It’s where I meet my people, keep up with what’s going on in mental health, and share what I’m learning. I’ve built a community of almost 50,000 followers, but I’ve noticed in the past few months, my reach has plummeted. I’m not an “influencer” professionally, but I do get my work that way, so I noticed. I have almost 50,000 followers, but the average post was suddenly getting only 2,000-4,000 views.

I was able to figure out when this dip occurred by pulling up my 46 best performing posts from the past year and sorting them by when they were published.

From November through April, I averaged almost five top-performing posts a month (29 posts across six months). That was my normal pattern.

Then from May through October, that average dropped to about one and a half top posts per month (9 posts across the next six months). Two of those months didn’t produce a single post that made the list.

Then came the experiment, when, in just one week, three posts broke into my top list, almost the same number I’d managed across the entire three months before it.

AND THAT’S NOT INCLUDING MY POST THAT WENT VIRAL YESTERDAY, though it did include the announcement of my experiment.

In my now-viral post, I got a lot of questions, and I wanted to try to answer those questions here.

I’m Not the First Person to See This

I should say up front: none of this started with me. Other women noticed these patterns long before I ran my little experiment. Cindy Gallop has been talking about gender bias on this platform, loudly and frequently, for a long time. She even ran her own test where she and a male colleague posted the exact same thing and his version took off while hers barely moved, even though she had 15x more followers.

There’s also actual research on this. A paper in the Journal of Behavioral Economics for Policy looked at gender code switching and found that women in male-dominated environments often shift their communication style to match whatever the system rewards. The visibility goes up, but the cost goes to the women who have to do the shifting.

A few weeks ago I mentioned declining reach on someone’s post (a man who teaches people how to build an audience on LinkedIn) and he said nothing had changed. He suggested my profile was stale and told me to try new formats, shake things up, expand my network. All reasonable suggestions, ones I tried with no results.

So I’m not the first person to see it, but I’m clearly not the last.

What Exactly Did I Change?

A lot of people asked for specifics, so here they are. I changed two things at the same time, and the third change rolled out over the course of the week.

1. I changed the gender field on my profile.

From female to male.

2. I changed the communication style of my LinkedIn Profile information.

I rewrote my headline, my About section, and several older posts in a more traditionally male-coded, agentic voice.

It was the same ideas and same topics, just a different tone.

To do this, I used the following prompt, along with this full article by Martyn Redstone on LinkedIn algorithmic gender bias:

> “I want to update my LinkedIn profile to make it more “Male Coded” based on this article. I will share my LinkedIn below, and I want you to make edits.”

(Then I pasted the text of the article.)

Here’s what the changes actually looked like.

Headline

Before:

“Mental Health Communicator. Founder of Therapy Trust Collective. Copywriter, Clinician, Consultant. Clinician Advocate.”

After:

“Founder, Therapy Trust Collective | Copywriter & Strategist for Mental Health Brands | Licensed Clinical Social Worker | Driving Ethical Growth in Behavioral Health”

About Section

Before:

“I started out as a licensed clinical social worker, doing school-based therapy. Then I entered the mental health tech space and saw how easily ethics get sidelined when money’s in the room. That realization changed my career.”

After:

“I’m a licensed clinical social worker turned strategist who builds and scales ethical mental health brands.”

Before:

“I help shape messaging that reflects real clinician experience, not just what looks good to investors.”

After:

“I now lead communications and strategy for companies that want to grow without compromising integrity.”

Before:

“I still bring a therapist’s lens, aka attuned to pain, trained in listening, and fluent in nuance- but I’m applying it to the larger conversation about work, money, and meaning.”

After:

“If you’re building a mental health company and need messaging that commands credibility, moves clinicians, and strengthens your market position, let’s talk.”

Posts

Before:

I love seeing AI platforms giving therapists more control over how far into their care processes AI should come.

You feel comfortable with AI, and you have a robust consenting process? Awesome, as long as your clients are cool with it, use AI transcription to help with documentation.

You only want AI to help you tidy up notes? Cool. No judgement.

I don’t think any platforms should take hard stances on these things, because clinicians are in charge of their own practices, and their own ethics, and at the end of the day, their own licenses.

Giving therapists in-platform control (that lets them increase AI use if and when they feel comfortable) is the WAY.

After:

I’m glad to see AI platforms giving therapists more control over how much AI is involved in their work.

If you’re confident in your process and have strong consent protocols, use transcription to streamline documentation. It saves time and increases accuracy.

If you prefer to keep AI at the edges of your workflow and only use it for note cleanup, that’s a valid approach too.

Platforms shouldn’t be dictating what “responsible use” looks like. Clinicians are accountable for their own practices, ethics, and licenses. That authority needs to stay with them.

The right move is building systems that let therapists adjust their level of AI integration over time. That’s how you build trust, adoption, and long-term value.

What Happened to My Numbers

Once I made the changes, the shift in reach was immediate. Here’s the simple breakdown:

In the two weeks around the experiment, I had 144,622 impressions.

In the seven days of the experiment, I had 116,251 impressions.

That’s a 415% increase from the week before.

The graphic below shows exactly where the experiment started. The spike lined up directly with the change in tone and the demographic update.

And just to rule out the possibility that I “just had a good week,” I didn’t use new content. I used recycled posts on purpose. At least one had performed well in the past, and at least one had done poorly. But none of the posts, with the exception of the announcement of my experiment, was new content.

Disclaimer: I Am Not a Scientist

It’s probably a good time to talk about the limitations of what I did, because this was not a scientific study and I’m aware of that. I follow a lot of scientists and researchers, and I love reading what they say about methodology, but my last research class was in 2014. I’m not pretending this was a peer-reviewed experiment.

For starters, I changed more than one thing at the same time. I switched my gender field, rewrote my profile, and started posting in a different tone. Three variables at once is not how you design a clean study, but it’s what I did. I also announced the experiment, which isn’t ideal, but I didn’t want people to think I’d completely changed personalities overnight. On the first couple of posts, I even left comments explaining the tone shift because I didn’t want to confuse my audience. That choice makes sense relationally, but it adds more noise to the data.

The timeframe was short, just one week. It’s enough to notice a pattern, but not enough to know how stable the pattern would be over time. And this is LinkedIn, after all. The algorithm changes constantly, audience behavior shifts, and there are always external factors affecting reach. There’s no way to isolate all of that.

Practically speaking, the reason I stopped after a week is because the results were so dramatic that I didn’t feel like continuing would tell me anything new. The pattern was already clear, and the experience was becoming more frustrating than informative.

Thoughts and Reflections

Reading the comments on the viral post was its own experience. I genuinely appreciate healthy skepticism. I try to practice it myself, and I welcome people who want to understand the details or ask for data. But there’s a difference between someone saying, “Can you show your methodology?” and someone saying, “This point might land if you actually showed the before/the data,” as if the entire thing existed to meet their personal standards.

Another clear pattern emerged. Every supportive comment from women was simply supportive. Not every man was negative, but every negative comment came from a man. That doesn’t make men the problem; it just means that if someone was going to dismiss or minimize the experiment, it was going to be a man. And some of the comments were almost absurd. One person called me a donkey. Another insisted that my logic was flawed because he uses a “male agentic voice” and doesn’t get views. His posting history consists of one post from ten years ago. When AI evaluated it and identified it as more communal and feminine in tone, he didn’t take it well.

Another man confidently explained that “men are good at some things and women are good at some things,” especially in business, and that this is simply how the world works. It’s a nice sentiment until you compare it to actual outcomes. Women receive only about fourteen percent of venture capital funding, but fourteen percent of startups that reach unicorn status have a woman co-founder. Women outperform the investment they’re given. The data doesn’t support the assumption that men are inherently better at business. What it does show is that there is a communication style people associate with business, and if you don’t use it, you’re assumed not to understand the terrain, even when results tell a very different story.

This is where the mental health piece comes back into focus. Some industries are built on collaborative language. Mental health is one of the biggest. Relational communication is literally the work. But business culture (and now LinkedIn’s algorithm) often treats relational language as less authoritative or less “professional.” That creates the same dynamic I see inside mental health companies all the time: business values and mental health values running into each other at full speed. Companies try to communicate with clinicians using language that fits a growth model, then assume clinicians are “resistant” when that language doesn’t land. Meanwhile, clinicians work in a field where the soft metrics actually determine the outcomes. The lesson being that you can’t act like soft metrics don’t matter when you’re building a business that quite literally is all about soft metrics.

When platforms reward only one communication style, it forces people in relational fields into a lose-lose. You have to choose between using language that gets visibility and using language that actually works in your industry. Either way, something gets compromised. And mental health gets hit especially hard because so much of our work depends on nuance, rapport, and context, and those are the things we tend to value in communication with not just our clients but also our colleagues.

And this puts me in a strange position professionally. I ghostwrite for people in the mental health space, including CEOs and founders who are trying to reach and build trust with therapists. Before this experiment, LinkedIn was a reliable place to do that. Now I’m not sure I can tell them that anymore. The tone therapists respond to—the tone that actually works in this field—is the tone the algorithm suppresses. It means the very people mental health leaders need to reach will never see the posts written in the voice that would resonate with them. If LinkedIn doesn’t care about the sexism piece, I hope they at least pay attention to this part, because it’s actively undermining an entire industry, and represents a risk to their business model.