Putting links in the first comment vs the post body: 2026 data

The habit is older than the data people now cite for it. Write the post, hide the URL under the first comment, and tell the reader the link is down there. In 2026 that move is no longer a free pass on distribution, and it is a worse place to put a URL you actually need clicked.

LinkedIn has not published a percent taken off a post because it contains a link. The studies that do publish a method, a sample, and a baseline disagree with the "60% and the comment is the same" columns, and they also disagree with each other on the size of the gap. What they agree on is narrower, and it is the part you can operate. Since mid-May 2026, a typed URL, a preview card, and a line that points at a link in the comments each sit below the same author's posts that have no link. The three placements are close. Hiding the URL does not buy the reach back.

The feed mechanics those studies sit on are in the LinkedIn algorithm guide. The actions LinkedIn has actually named as distribution limits are in algorithm penalties. This page stays on one decision: body, first comment, or no URL in the feed post at all.

What LinkedIn has published

Oposto's 6 October 2026 review of the public help pages is the clean official reading. The help page on how the feed ranks content lists context, profile, network, and activity. It does not list outbound links. The March 2026 feed notes target repetitive, low-substance posts, engagement bait, and comment automation. Hristo Danchev's 12 March 2026 engineering post describes retrieval and a ranker that predicts clicks, skips, long dwells, likes, comments, and shares. It does not describe a link detector.

One help note does constrain the composer. On the page for sharing video, LinkedIn says you can share either a URL or an image in a post, not both at the same time, and that a URL gets a preview image taken from the site being shared (LinkedIn Help). A post is also capped at 3,000 characters (LinkedIn Help). Neither page says the preview, or the bare URL, reduces who sees the post.

Flypost's 17 October 2026 write-up of its own sample reports a comment from Daniel Roth, LinkedIn's editor in chief, on Amanda Natividad's post about that study: "we don't have anything in the algo about links." Flypost's paraphrase of his explanation is that people who click are off reading, so they do not react, and the post then has fewer of the signals used to decide who else sees it. That is a reported comment, not a help-center rule, and Flypost counted reactions rather than impressions. It matches a mechanism you can test. It is not a published 60% tax.

Where the 60% figure comes from

Jodie Cook's Forbes columns on 23 July 2026 and 30 July 2026 state that an outbound link cuts reach by about 60%, and that putting the link in the first comment now costs the same. The text above the subscription wall does not show a post count, a baseline, or a date range. The same author's 12 January 2026 column said the opposite habit was outdated, and that a link could sit in the body without a significant penalty. Three columns, one year, no shared method. Treat the 60% as a claim that spread, not as a LinkedIn measurement.

MyFeedIn's September 2026 check says the widely shared 60% traces to third-party blogs, not to LinkedIn. On 40,785 posts from 599 people, September 2025 through September 2026, each post scored against its author's own median impressions, a post with a link in the text landed at 0.93. That is about 7% fewer impressions than that author's typical post. Their instruction is explicit: if the link is the point, write the idea in the text, and do not contort the post into "link in the first comment" to save 7%.

The 2026 numbers, on the baseline each study actually used

Raw medians and same-author gaps answer different questions. A raw median mixes a 400-follower account with a 100,000-follower account, and it mixes a link card (the article format) with text, image, and video. A same-author gap asks what happened to this person's posts when a link was present, against this person's posts without one. Use the second when you decide where your URL goes.

MagicPost's June 2026 cut of 566,957 posts with synced analytics, the 12 months up to June 2026, reshares and deleted posts excluded, medians:

In that pooled cut, the body URL did not trail link-free posts. The card did: 414 versus 795 median impressions for the broader "no attached link" bucket of 534,030 posts. MagicPost flags the confound. An attached card is the article format, and article posts were the weakest format in the same pool (414 impressions, against 664 for text-only with no link). Follower bands still showed the card near half: 455 to 266 under 5,000 followers, 1,401 to 726 from 5,000 to 50,000, 9,140 to 3,705 above 50,000. The share of posts carrying an attached link fell from 12.2% in 2022 (65,739 posts) to 5.6% in 2026 (468,845 posts). Creators already walked away from the card. That walk does not prove a rule.

The same-author cut is the one that speaks to 2026 placement. It lives in MagicPost's algorithm-changes study of 687,000 posts from 16,000 creators, June 2025 to September 2026. Each link row is against that author's posts without links. "Before" is before mid-May 2026. "Since" is after that break.

Read the row, not the folklore. The card got less expensive on this baseline (39% to 19%). The typed URL got more expensive (12% to 21%). The "link in the comments" line also got more expensive (8% to 16%). Since mid-May, the three placements sit in a five-point band. The comment is not the shelter.

A second MagicPost study, 684,000 posts, measured the sentence that points at the comment rather than the URL inside it. Posts that tell the reader the link is in the comments sat about 11% below that author's average post, both before and after late July 2026. Posts that end on "Comment KEYWORD" fell to about half. The algorithm-changes study dates the sharper break to 27 July 2026: on the same 1,983 creators, all posts were about 2% lower that week, while "Comment KEYWORD" went from 10% above the author's average post to 47% below, and other comment requests ("in the comments", a downward arrow, "let me know") went from 2% below to 19% below. A "link in the comments" closer is in that second family. You pay for the sentence even when the URL itself is one scroll away.

MyFeedIn's 7% and MagicPost's 21% are not a contradiction you have to resolve by picking a winner. Different windows, different link definitions, different baselines (median impressions versus a log-scale mean of nearby posts). Both are below the 60% column. Neither is a LinkedIn rule.

Reach is not the click

A placement that protects impressions can still fail the job. Alan S.'s 25 June 2026 note summarizes Richard van der Blom's paid Algorithm Insights report. The method is not on that public page, so treat the figures as a secondhand commercial summary, not as an open dataset.

On the older cut he describes, an outbound link in the post got about 40% less reach than a post without one, and it converted impressions best. The same link in the comments kept more reach and lost up to 75% of the conversion. Editing the link in after publishing sat in between: about 25% less reach and 40% less conversion. His June 2026 update of the same report says LinkedIn was then reducing reach on posts that hide an external link in the comments, by as much as 80% of what that post's organic reach would have been when the comment generates a preview, and that a comment link performed 3.4 times worse than the same link in the body. He also reports more than one outbound link costing about 9% of reach against about 18.8% for a single link, when the links add context rather than being the whole post. Buy the report if you need the tables. Do not run the account on a paraphrase. The direction is the part that matches the open studies: the comment is where the click goes to die, and in 2026 it is no longer an obvious reach win either.

Three product facts make the comment a bad default even before any percent:

• The reader who does not open comments never sees the URL. A reshare of the post does not carry your comment with it.
• The preview card LinkedIn builds is for a URL in the post. A comment is a different surface, and on a busy thread it is not reliably "first" for every reader.
• The sentence "link in the comments" is the object the July 2026 samples mark down. You can lose the impressions and still have the URL buried.

Flypost's reaction count is the other half of the trade. A click takes the reader off the post, so likes and comments fall even when the visit to your site was the point. Judge a link post on clicks and on qualified visits, not only on impressions. A post with no URL can win the impression column and send nobody to the page you needed them to read.

A workflow for one personal profile

Boykli publishes only to the personal profile you connect with LinkedIn OAuth. The scopes are openid, profile, email, and w_member_social. The queue sends the approved text. It does not open the thread, post a comment, like, or message. A company Page is a different analytics surface, and this product does not queue to it. If the URL has to be a comment, you type that comment yourself after the post is live.

Do this in order. Change the placement, and hold the other variables still.

1. Name the job before you pick a slot

Three jobs, three placements.

If the long version lives on your domain, the feed post is a teaser with the decision in it, and the canonical URL stays on the page you want search engines to keep. That split is LinkedIn newsletter versus a blog. A newsletter edition can carry an inline link. A feed post is the surface this page is about.

2. Write the point so the post survives without a tap

Open with the number, the refusal, or the sequence. Put the URL in the last third, as a normal sentence, once. Do not open with the URL, do not stack three, and do not close with "comment the word and I will send it." The July samples are harsh on that closer, and the post that withholds the point is the failure mode in engagement versus value.

If the post also needs a picture, pick one. LinkedIn's help note is a composer rule: a URL or an image, not both. A document post is a third object. Links drawn inside a PDF are not the preview-card case, and they are not a tap target the way a body URL is. The document path is the carousel workflow, not a way to smuggle a link past the feed.

3. Preview the line, then queue one clock

Read the draft at phone width and desktop width before it leaves. A URL that wraps into the see-more fold is a body link the reader still has to hunt. That check is LinkedIn post preview. The queue on the scheduling tool stores the weekdays you toggle, one time of day, and one timezone. It stores the text you approved. It does not store a comment to fire two minutes later.

4. If you still comment, do it by hand, once, without a gate

After the post is live, add one comment from the same account if the URL was not in the body. State what the page is. Do not ask for a keyword, a like, or a repost in exchange. Do not edit the live post twenty minutes later to paste the URL in as a workaround. Alan S.'s summary of the older van der Blom cut already had that edit losing both reach and conversion relative to putting the link in at publish. The penalties page is the right place for the rumor that an edit zeros impressions. That zero is not in the 2026 notes. The edit is still a muddy test, because you changed the object after some people had already seen it.

5. Score the next six posts against your own baseline

Keep format, length band, and weekday fixed. Alternate three posts with one body URL and three with no URL and no "link in comments" line. The next day, copy impressions, members reached, and the in-network versus out-of-network split. LinkedIn's combined analytics graph is described as complete through yesterday, so ignore a post from this morning. Add a column the analytics screen will not compute: clicks, from the destination you control. A 15% impression dip with a real signup is a successful link post. A comment link with a normal impression count and three clicks is a failed one.

Six posts is a reading, not a law. MyFeedIn kept authors with at least ten posts before scoring a link. Your account will not recreate a 687,000-post study. It will tell you whether your buyers tap. Stop the test if a restriction notice appears. The queue is not the cause of that notice. Software that acts in the LinkedIn tab is a different class, covered on the penalties page.

Boykli can hold the draft and the clock. Create posts writes from a note you supply. Voice Agent may use only the transcript of an interview you sat. Neither step invents a link strategy, and neither posts the comment. Managed caps are 210 posts, 150 carousels, 150 images, 25 Voice Agent sessions, and 130 post refinements a month. BYOK uses your OpenRouter key for chat and images and your Google AI Studio key for Voice Agent, with no Boykli generation cap. You pay those providers for usage. The product is on boykli.com.

What to stop doing this week

Stop announcing the comment. If the last eight posts end with "link in comments" or a keyword, the July 2026 break is the first thing to remove, before you debate cards versus bare URLs. Keep one URL, in the body, on the posts whose job is the click. Leave the URL out when the post is the whole point. Read clicks beside impressions. The 60% column is not an operating number. The same-author band of roughly 16% to 21% since mid-May, on MagicPost's sample, is the size of bet you are making, and it is similar for every placement they measured.

FAQ

Does LinkedIn penalize external links in 2026?

LinkedIn has not published a penalty percent for a URL in a post. Danchev's March 2026 engineering note and the help page on feed ranking do not name one. Third-party samples do show a gap. MyFeedIn measured about 7% fewer impressions for a link in the text, against the author's own median, on 40,785 posts. MagicPost measured a typed URL at 21% below the author's posts without links since mid-May 2026, on the 687,000-post study. Forbes columns in late July 2026 state about 60% and do not show a sample in the public opening. Use the studies that show the method. Do not treat any of them as a switch LinkedIn documented.

Is the first comment still safer than the post body?

Not on the 2026 same-author numbers. MagicPost has a "link in the comments" line at 16% below the author's link-free posts since mid-May 2026, against 21% for a typed URL and 19% for a card. Posts that tell the reader to look in the comments sat about 11% below that author's average post in the 684,000-post comment study, before and after late July. The sentence is a cost. The click is worse in the comment than in the body, on the secondhand van der Blom summary Alan S. published on 25 June 2026. Put the URL in the body when you need the tap. Leave it out when you do not.

Does a preview card cost more than a typed URL?

In MagicPost's pooled medians up to June 2026, yes: 414 median impressions for a card, 858 for a URL in the body, 786 with no link. Part of that is format. The article post is the card. On the same-author baseline since mid-May 2026, the card (19% below) and the typed URL (21% below) are close. The card used to be the expensive one (39% before mid-May). It is not a free preview, and it is no longer in a different league from typing the URL.

Can a LinkedIn post contain both a link and an image?

LinkedIn's help note on sharing video says you share either a URL or an image, not both. A URL gets a preview taken from the site. If the image is the post, keep the URL out of that post. If the URL is the post, skip the image upload.

Will Boykli post my link as the first comment?

No. The queue publishes the text you approved to the personal profile. There is no comment step in the product. A URL you put in the draft is in the body. A URL you wanted under the post is something you add on LinkedIn after it is live.

Should I publish without the link and edit it in later?

That tactic is a 2020-era habit. Alan S.'s summary of the older van der Blom cut had the edit losing reach and conversion versus including the link at publish. Editing also changes the post after some impressions have already landed, so the before-and-after on one post is not a clean test. Decide the placement before you queue it.

How many links belong in one post?

One, if the click is the job. Alan S. reports van der Blom's paid update at about 9% reach impact for more than one contextual link, against about 18.8% for a single link. That is a commercial summary without a public method. It is not a reason to stack URLs. A second link belongs in a later post with its own point, or on the canonical page. The reason the sentence has to come from work you did is the constraint in building Boykli.