The question sounds dramatic, but creators who publish consistently on LinkedIn eventually run into the same friction point : posts with links seem to perform worse. It’s become almost folk wisdom. Experienced users have internalized the rule. Newer users are taught it. And creators who rely on LinkedIn for distribution quietly accept the reality that every outbound link feels like a tax.
But folk wisdom isn’t data. And platform behaviour doesn’t always match the tidy hypotheses we come up with to explain it. So when a Substack Note from writer JHong pointed out that “link in comments” has become basically invisible — and that including the link in the main post was the only real way to get clicks — it raised a more interesting question. If link placement matters, does link destination matter too?
For someone publishing simultaneously on Substack and a LinkedIn Newsletter, that question isn’t academic. It affects distribution, visibility, strategy, and growth. And it’s a question worth testing instead of speculating.
Why LinkedIn Might Penalize Links in the First Place
To understand any modern feed algorithm, you need to understand how the business behind it makes money. LinkedIn is, above everything else, an advertising platform. Yes, it sells subscriptions. Yes, it offers hiring tools and premium features. But the bulk of its revenue still comes from selling targeted ad impressions. That means one metric matters more than anything else: time spent on-site.
Any link that takes someone away from the feed interrupts the behaviour LinkedIn is optimized to monetize. So if the platform deprioritizes posts with links — especially links to other publishing platforms like Substack — it’s acting rationally within its incentives. A link is a potential leak in the revenue model.
But not all links are equal, which is where the real question emerges. Does LinkedIn treat a link to its own ecosystem (a LinkedIn Newsletter article) differently than a link that directs a user to a completely external platform like Substack?
I designed a two-week experiment to find out.
How the Experiment Worked
For fourteen days, three posts were published each day :
- A control post with no link at all.
- A post linking to the LinkedIn-hosted version of an article.
- A post linking to the Substack-hosted version of the same article.
The posts were kept as similar as possible : same topics, same visual structure, similar tone, and the same posting window each day. The idea wasn’t to measure clicks or conversions, that wasn’t the concern. The only question was : Which posts does LinkedIn show to people?
The metric of interest was impressions — the number of times each post appeared in a user’s feed.
The Results, and Why They’re Interesting
If the intuition behind link penalties is correct, the control posts should perform best. And they did, but the scale of the difference was surprising.
Across the test period :
- Control posts generated 140% more impressions than posts linking to LinkedIn Newsletter articles.
- LinkedIn-linked posts only slightly outperformed Substack-linked posts (493 vs. 436 impressions).
- The gap between the two link types was small enough that it’s hard to claim any real advantage for on-site links.
What surprised me most wasn’t that links depress reach, but that LinkedIn’s own Newsletter feature didn’t receive preferential treatment. If anything, the platform treated both link types similarly, suggesting that the act of clicking itself — on-site or off-site — interrupts the feed enough to trigger lower distribution.
From the platform’s perspective, it makes sense. Even an on-site click disrupts scrolling behaviour, which weakens the core ad-driven engine. The feed is the product. Anything that pulls a user out of that rhythm becomes a liability.
Behavioural Trends Worth Noting
Beyond link performance, the dataset surfaced a few patterns that matter for creators :
- Wednesdays performed best, generating 24% more impressions than any other day.
- Morning posts outperformed afternoon posts by about 17%.
- Posts with two or more reactions tended to accelerate, confirming the widely observed snowball effect of engagement.
None of these patterns should be generalized too aggressively… audiences differ, routines differ, and LinkedIn’s algorithm is dynamic. But the trends are directionally useful and reinforce an important reality : most distribution problems are behavioural, not algorithmic.
What Builders Should Take Away From This
If your primary goal is reach, removing links is the simplest way to improve post visibility. If your goal is traffic, then links are necessary, but you should rely on them strategically rather than by default.
The real value of this experiment isn’t the specific numbers. It’s the reminder that attention platforms optimize for themselves, not for creators. And if you want to build an audience that compounds, you’re better off understanding the underlying incentives than guessing at the shadows cast by the algorithm.
This article is a part of my series on topics for entrepreneurs, intrapreneurs, and people who just love building things. I podcast and post weekly with tools and guides on The Journey. Check out the companion piece here : https://6catalysts.substack.com/p/does-linkedin-hate-substack