I've been posting on LinkedIn every weekday since late 2025. Seventy-four posts. Some hit 107,000 impressions. Some hit 575. Same author, same audience, same posting time, same platform.
For months I thought the difference was topic selection. Commerce posts outperformed leadership posts. Contrarian takes outperformed tactical frameworks. Posts about Shopify and Salesforce outperformed posts about AI tooling. So I built a content strategy around topics. Fifty percent commerce. Thirty percent contrarian takes. Twenty percent tactical frameworks. I scheduled posts by category. I tracked which topics hit and which ones missed.
It worked ... sort of. My floor improved. My worst posts stopped being catastrophically bad. But the ceiling stayed unpredictable. Some commerce posts hit 107,000 impressions. Others hit 575. Same category. Same level of research. Same posting time. Same author with the same credentials writing about the same industry.
The topic theory was wrong.
What the Data Actually Shows
I pulled every post, sorted by impressions, and looked at what the top performers had in common. Not by topic. By what the post did to the reader.
Every post above 5,000 impressions threatened the reader's identity.
Every post below 3,000 impressions gave the reader something useful.
That's the entire finding. The topic, the format, the time of day, the day of week ... none of it mattered as much as one question: did the reader see something about themselves they didn't want to see?
Here's what that looks like in the data.
My top post, 107,328 impressions, opened with "Shopify lost the enterprise war." I'd spent a decade on Salesforce Commerce Cloud. I'd publicly questioned whether Shopify could compete. Then I wrote a post saying the war Shopify lost didn't matter anymore because they'd already won the next one. Both sides of the platform debate felt something. Shopify people felt validated. SFCC people felt challenged. Everyone saw their own position reflected back at them in a way that was uncomfortable.
My second biggest post, 62,819 impressions, opened with "The worst advice I got as a new leader: Managers shouldn't write code." Every engineering manager who stopped coding felt exposed. Every manager who kept coding felt vindicated. Both groups engaged because the post held up a mirror to a choice they'd made about their own identity.
My third, 24,033 impressions: "I've seen the best engineers hit ceilings hard. Not because of their code. Because they left PR comments like 'did you even try?'" Every senior engineer reading that thought about their own comments, their own behavior in reviews, their own reputation with teammates.
Now look at the posts that didn't break through.
"Here's how to measure technical debt impact." 1,136 impressions. Useful. Actionable. Nobody felt anything reading it. They nodded and scrolled. "Tie your work to business value." 709 impressions. I tried this message three times with three different hooks. Marriage analogy. Direct framework. Question format. Three attempts: 620, 1,145, 709. The audience agreed with the advice every time and scrolled past every time. Agreement doesn't generate engagement. Discomfort does.
"Agentic commerce won't cannibalize your revenue. It'll give it to your competitor instead." 575 impressions. Analytically sound. Well-researched. Nobody's identity was threatened by a prediction about AI agents. The pattern held across all 74 posts. Thirteen posts broke 5,000 impressions. Every single one of them made the reader question something about themselves ... their technical depth, their leadership approach, their platform loyalty, their career trajectory. The posts that taught without threatening landed between 1,000 and 3,000. Consistently.
The Same Topic, Two Different Results
The clearest proof is when the same topic produces wildly different results depending on whether the mirror is present. I write about commerce platforms regularly. Commerce posts with a strong personal opinion and an identity threat averaged over 15,000 impressions across my top performers. Commerce posts that analyzed trends without threatening anyone averaged under 1,500.
Same author. Same expertise. Same audience. Same topic.
"Shopify lost the enterprise war" at 107,328 held a mirror. It told SFCC engineers that the platform war they'd been winning might not matter anymore. It told Shopify skeptics they might be wrong.
"Agentic commerce won't cannibalize your revenue" at 575 handed out analysis. It told the reader something interesting about the future. Nobody felt uncomfortable. The difference between 107,000 and 575 within the same content category is whether you put yourself and your reader on the line.
Why Mirrors Work and Worksheets Don't
LinkedIn's algorithm rewards engagement. Comments, shares, reactions, dwell time. That's the mechanical explanation. But the psychological one is more interesting.
When someone reads a useful framework, they process it cognitively. They evaluate whether it's correct. They might bookmark it. But they don't feel compelled to respond because the post didn't challenge anything about who they are. It just added to what they know.
When someone reads a post that threatens their identity ... "you think you're still technical but you're not" or "the skills that made you senior are keeping you stuck" ... they can't just process it cognitively. They have to respond emotionally. They either agree (and that agreement costs them something because it means admitting the uncomfortable thing) or they disagree (and that disagreement is passionate because their self-image is at stake).
Both reactions produce engagement. Agreement produces thoughtful comments about personal experience. Disagreement produces passionate pushback. Both feed the algorithm. Both generate reach.
Useful content gets respect. Identity threats get reactions.
What I Changed
I rewrote my entire content strategy around this principle. Instead of organizing posts by topic ... commerce, leadership, frameworks ... I organize them around one question: what does my audience believe about themselves that isn't true?
The topic is just the surface the mirror reflects off of.
A post about technical debt becomes: "You keep calling it technical debt because admitting it's just code you don't like would mean your last three refactoring projects were vanity work."
A post about AI adoption becomes: "The engineers on your team who refuse AI aren't protecting quality. They're protecting the only thing that made them feel special."
A post about platform strategy becomes: "Your entire commerce team is about to be exposed for the product data mess they've been hiding behind a pretty storefront for five years."
Same topics I was writing about before. Different entry point. The entry point is always the reader's self-image.
The Uncomfortable Part
Here's where this essay holds a mirror at me.
I spent months optimizing for the wrong variable. I tracked topics, tested hooks, analyzed posting times, built content calendars with percentage allocations. I was doing what most people do with content strategy ... treating it as a scheduling problem when it's actually a psychology problem.
The data was always there. I just wasn't looking at it the right way because I was too focused on the mechanics of what I was posting and not focused enough on what my posts were doing to the people reading them.
Most engineering leaders posting on LinkedIn are making the same mistake right now. They're sharing frameworks, writing tutorials, giving advice, and wondering why their posts don't break through. The content is good. The expertise is real. The advice is sound.
Nobody cares.
Not because the advice is wrong. Because the advice doesn't make them feel anything. It adds to their knowledge without challenging their identity. And knowledge alone doesn't generate the kind of engagement that makes content travel.
What This Means for You
If you're an engineering leader building a presence on LinkedIn, you probably have useful things to say. You've solved real problems. You've learned hard lessons. You have frameworks that actually work.
None of that matters if you're handing out worksheets.
Your audience doesn't need more information. They have more information than they know what to do with. What they need, and what they'll engage with, is someone willing to say the uncomfortable thing they've been thinking but haven't had the courage to say out loud.
The post that changes your reach is the one where you hold up a mirror and let the reader decide what to do with what they see.
Every piece of content you create, ask yourself one question: will this make someone uncomfortable about who they are right now?
If the answer is no, you're writing a worksheet. And worksheets don't travel.