Do external links kill your LinkedIn reach? (-36%)

Across 358,021 LinkedIn posts analyzed, those with an external link show a median of 448 impressions versus 705 for posts without one. That's -36%. Here's the raw number — and here's what it actually means.

In short

→ Posts with an external link: median 448 impressions (74,891 posts) → Posts without a link: median 705 impressions (284,130 posts) → Gap: -36% median reach in favor of link-free posts → Posts with links collect slightly more likes (probable selection bias) → This is an observational correlation, not a confirmed direct penalty

What the data actually shows

From the fresh corpus §13.8 of our study on 516,144 LinkedIn posts:

| Condition | Number of posts | Median impressions | |-----------|-----------------|-------------------| | With external link | 74,891 | 448 | | Without external link | 284,130 | 705 |

The gap is clear and statistically robust. Posts without links have 57% more median impressions than posts with links. This isn't noise — it's a reproducible pattern across nearly 360,000 posts.

LinkedIn penalty or selection bias?

That's the real question. Two readings coexist — and they're not mutually exclusive.

Reading 1: LinkedIn deprioritizes posts with outbound links. LinkedIn has every incentive to keep users on the platform. An external link is an exit door. Dwell time — the time spent on a post — is one of the algorithm's strongest signals. If people click and leave the platform immediately, the post loses ranking.

Reading 2: Posts with links are, on average, less polished. Creators who post links are often in "promotion" mode — sharing an article, a sales page, a newsletter sign-up. These posts statistically have weaker hooks and less engaging structures. The link may not be the cause of low reach — it may be a symptom of less developed writing.

Our data can't definitively settle this. It measures a correlation, not causation. What is certain: the -36% gap is there, and it's large enough to take seriously.

What high-reach creators actually do

The standard practice: put the link in the first comment.

Publish the post with no link in the body. Immediately after publishing, reply to your own post as the very first comment with the URL. Readers who want the link find it effortlessly. The algorithm indexes the post as link-free content in the feed.

This isn't an obscure trick — it's a publishing hygiene adopted by the majority of creators with significant reach. It doesn't guarantee +36% reach (reminder: correlation, not causation), but it eliminates the most obvious risk factor.

The same logic applies to hashtags: our analysis of LinkedIn hashtags shows their impact on reach is far weaker than most people think — and that 4 or more hashtags actually hurt engagement.

When a link in the body can still hold

There are cases where a link in the body doesn't destroy your post:

→ Your audience is highly engaged and clicks consistently (positive signal for the algorithm) → Your hook is strong enough to generate dwell time before people click → You're sharing a unique resource your subscribers have been genuinely waiting for

And cases where a link in the body clearly costs you:

→ Direct promotion post with the link pushed front and center in the opening lines → Third-party article shares with no worked hook ("here's a great article") → A shortened URL pointing to an unknown destination

What to do in practice

1. By default, put the link in the first comment — reply to your own post immediately after publishing.
2. Test both versions on your account: same content, once with the link in the body, once in the comment. Measure the gap on your own history.
3. Don't sacrifice post quality to avoid a link: a weak post without a link is still a weak post. Hook, structure, and length weigh far more than the presence of a link.
4. Analyze your past posts to see whether the correlation holds for your specific audience.

Before you publish, run your post through the free LinkPost analyzer. In 30 seconds you see where your post falls short, what can compensate for an external link, and how to adjust without rewriting from scratch.

Observational study by LinkPost, corpus §13.8, 358,021 posts analyzed across 30,843 profiles (2020 to April 2026, 62% French-language content). Findings are correlations. Sampling bias is possible. Full methodology and limitations in the playbook.