The LinkedIn call to action: how to end a post so it earns comments instead of silence

The strongest LinkedIn call to action is a question narrow enough that a reader can answer it from their own experience without thinking hard. The weakest is “Thoughts?”, because it hands the reader the hardest part of commenting: deciding what to say. In our analysis of 12,988 posts, comments are the metric with the widest gap between strong and weak posts. The top decile carries a median of 72 comments against 5 in the bottom half, a 14x gap, while reactions differ by roughly 3x. Whatever the ending of your post does, it does it to that number.

What is a LinkedIn call to action and where does it go?

A LinkedIn call to action is the last line or two of a post, and it does one job: tell the reader what happens next. That can be a question, an invitation, a request, or an explicit deliberate silence. It sits after the body and before your hashtags, if you use any, and it is the only part of the post the reader sees immediately before deciding whether to do anything.

Structurally, a LinkedIn post has three load-bearing parts. The hook, which is everything before the fold at roughly 210 characters and decides whether the post is read at all. The body, which decides whether the reader finishes. And the ending, which decides what they do about it. Most writing advice covers the first, some covers the second, and almost none covers the third, which is odd given that it is the part that produces the metric the platform weighs most.

We should be honest about a limit in our own data here. Our dataset captures each post's text only up to LinkedIn's “see more” fold, which means we can see hooks and we cannot see endings. So we cannot tell you that a particular CTA phrasing produced a particular comment count. What we can tell you is the shape of the prize: how large the comment gap is between strong and weak posts, and which formats and topics sit on which side of it. The judgments about phrasing in this article are reasoning from that, plus editorial experience, and we will flag which is which.

Why are comments the metric your ending should target?

Because comments separate posts far more sharply than reactions do, and because they cost more to earn.

From 12,988 posts by 65 creators. Correlational, not causal, and the cohort skews toward established accounts.

Read the ratio rather than the values. Between the top decile and the bottom half, reactions roughly triple. Comments increase roughly fourteenfold. A reaction is a tap. A comment means someone stopped, composed a sentence, and attached their name and job title to it in front of their own network. That is a much higher bar, which is exactly why it discriminates so well between posts that landed and posts that did not.

There is external support for treating comments as the higher-value signal. Hootsuite, citing an AuthoredUp analysis of over three million posts, reports that a save drives roughly five times the reach of a like and about twice the reach of a comment, which puts comments comfortably above likes in reach impact. AuthoredUp separately reports that posts drawing threaded discussion can see up to 2.4x the reach of ordinary posts. Both are on AuthoredUp's algorithm guide.

Format matters here too, and it is measurable in our data. The median native video post earned 27 comments (1,613 posts, 46 authors) and the median text post earned 21 (2,729 posts, 54 authors), while the median shared-article post earned 7 (3,813 posts, 60 authors). If you want comments, the format you choose narrows the ceiling before your CTA gets a chance to matter.

Why is “Thoughts?” a weak call to action?

Because it is three questions pretending to be one. When a reader hits “Thoughts?” they have to decide which part of your post to respond to, what position they hold on it, and whether that position is worth publishing under their real name where their boss and their clients will see it. Most people stall on the first step and scroll.

Compare that to a question that has already done the first two steps: “What was the first thing you cut when your budget got halved?” The reader does not choose a topic, does not construct a position, and does not risk much by answering. They just remember what happened and type it. The comment writes itself, which is the entire design goal.

The same problem afflicts a family of endings that all look like CTAs and function as punctuation:

• “Agree?” A yes-or-no question generates yes-or-no comments, which is the shallowest form of engagement available.
• “What do you think?”Same defect as “Thoughts?” with more words.
• “Let me know in the comments!” Tells the reader where to put a response without telling them what the response is.
• “Repost if you agree.” Asks for a much larger commitment than a comment while offering the reader nothing.
• “Follow me for more content like this.” Asks for an ongoing relationship at the exact moment the reader has one data point.

A useful test: could this ending sit at the bottom of any post in your industry? If yes, it is generic, and generic endings signal that the writer had no particular reader in mind. That signal is legible, in the same way a generic hook is. The parallel with openers is exact, and the opening data shows what specificity buys at the other end of the post.

What are the four ways to end a LinkedIn post?

Every ending is one of four things. Choosing deliberately is most of the work.

A reasonable mix over a month of sixteen posts: eight questions, four with no CTA, three soft CTAs and one direct ask. The exact numbers matter less than the ratio. When direct asks exceed roughly one post in six, the whole feed starts reading as a sales channel and the other endings stop working too. Slot the ratio into your content calendar so it is decided in advance rather than in the moment.

How do you write a question people can actually answer?

Five rules. Each one removes work from the reader.

1. Ask about experience, not opinion.“What happened when you tried it?” is easier to answer than “Do you think it works?” People are more willing to report than to argue, especially publicly.
2. Narrow it until it is almost too specific.“How do you handle difficult clients?” is a survey. “What is the sentence you use when a client asks for something out of scope?” is answerable in ten seconds.
3. Make sure the answer is short. If a good answer requires three paragraphs, most people will not start. Design for one sentence.
4. Do not ask something you already answered.If the post laid out your full position and then asks for the reader's, you have left them nothing to add except agreement.
5. Answer it yourself first, in the post. Modelling the answer shows the length and register you are inviting, and it lowers the risk for the first person to comment.

There is a mechanical point too: question hooks appear in 19.0% of top-decile posts against 14.2% of bottom-half posts in our cohort, and posts that draw heavy comment volume over-index further. That measures openings rather than endings, but it suggests that a genuine question, wherever it sits, correlates with the posts people respond to. More on question construction in the hooks guide.

When should you use no call to action at all?

More often than most advice allows. Four situations where the ask actively costs you.

• After a personal story that landed.A post about something difficult that ends with “What about you?” converts a moment into a prompt, and readers feel the switch. Let the last line of the story be the last line.
• After a strong opinion.A clear position invites disagreement on its own. Adding “Agree?” makes the position look tentative, which undercuts the thing that made it worth reading.
• When the post is a resource. If you have given away a checklist or a framework, the value is the giving. An ask attached to a gift reframes it as a trade.
• When you have asked for something in each of the last three posts.Fatigue is real. A feed of endings that all want something teaches people to stop reading to the end.

The test is whether the post is complete. If a reader finishing it would naturally have something to add, the question helps. If they would naturally sit with it for a second, the question interrupts.

How does your CTA change whether you get comments or clicks?

These two goals pull against each other and you cannot serve both in one post.

A comment keeps the reader on the post. A click removes them from it. Everything about LinkedIn distribution favors the first: dwell time is a signal LinkedIn measures, per its engineering blog as reported by Hootsuite, and a reader who leaves cannot generate any of it.

Our data shows the format-level version of this tension clearly. Shared article and link posts make up 32.7% of the bottom half of our cohort against 22.1% of the top decile, over-indexing about 1.5x among weak posts, and they carry the lowest median comment count of any format we measured at 7. That does not prove links are penalized, and LinkedIn has denied a penalty exists. It does mean that in practice, posts built around sending people away perform worse on the metrics the feed rewards. The contradictory evidence is laid out in our external-links analysis.

The practical resolution is a split. Most of your posts end with a question and stay self-contained. The occasional post that genuinely needs a click puts the link in the first comment and uses the post ending to say where it is. You accept slightly worse reach on that post in exchange for the click, deliberately, rather than accidentally doing it every time.

What does a soft CTA look like in practice?

A soft CTA asks for something smaller than a meeting and larger than a like. It is the most underused ending on LinkedIn, and it is the one that produces inbound.

The mechanism is that it lets someone identify themselves without committing to anything. “If you are the person rebuilding a pricing page this quarter, reply and I will send you the questions I ask” costs the reader one sentence and surfaces exactly the people you want to talk to. “Book a call” costs them a decision about whether they want to be sold to, which most people resolve by scrolling.

Four soft CTAs that work, described generically because we cannot measure their performance in our dataset:

• The conditional offer.“If this is your problem, reply and I will send X.” Self-selecting and specific.
• The save prompt.“Worth saving if you are hiring this quarter.” Saves carry weight, per the AuthoredUp figures above, and this asks for one without being about you.
• The tag invitation.“Tag the person who taught you this.” Works only when the post genuinely involves other people; otherwise it reads as growth hacking.
• The continuation.“There is a longer version of this. Say so and I will write it.” Produces comments and your next post at once.

What does a post that maximises comments look like?

Here is a real post from our dataset. It is a document post about remote work, and its engagement profile is the inverse of a normal post.

The median post in our cohort earned 122 reactions and 12 comments, roughly ten reactions per comment. This one inverted that ratio by a factor of forty. The reason is visible in the text: the post asked people to vote in the comments, so commenting was not a response to the post, it was the point of the post. It then paid the commenters back by tallying the results and publishing them.

Two transferable lessons. First, an ask that makes the comment section the content will beat an ask that treats it as an afterthought. Second, the author returned to the post and rewarded participation, which is the part almost nobody copies. Replying inside the first hour matters more than the CTA phrasing does, and both matter more than the hour you published at. The mechanics are covered in the comments playbook.

Should your call to action change by post format?

Yes, because each format leaves the reader in a different state, and the ending should meet them where the format put them.

• Text posts end where the reader is already reading, so a question costs nothing to answer. This is the format where a question ending works best, and it is reflected in the numbers: the median text post in our cohort earned 21 comments against 10 for the median image post.
• Native video creates a gap between where the content ends (in the video) and where the response happens (in the caption). Put the question in the caption as well as saying it on camera, because a meaningful share of viewers watch without sound.
• Document carousels end on a final slide, which is the single best place on LinkedIn for a soft CTA. The reader has swiped through ten panels and has already demonstrated interest. Ask for the save or the reply on the last slide and repeat it in the caption.
• Pollshave the vote as their built-in CTA, which is why the caption ending should ask for the reason behind the vote instead. “Voted other? Say which” converts a click into a comment.
• Image posts are the hardest, because an image invites a reaction and not much else. If comments are the goal, the caption has to carry a question the image raises rather than one it answers.
• Link posts should not end with a second link. If the post already points somewhere, the ending should give the reader a reason to comment before they go, or the post gets neither.

Format choice sets the ceiling and the CTA decides how close you get to it. Since native video accounts for 26.3% of top-decile posts against 10.1% of the bottom half in our cohort, the format decision is worth making before the ending decision. There is more on that in the native video breakdown.

Do follow, connect and DM calls to action work?

They work in a narrow set of conditions and fail loudly outside it, which is why they show up in so much bad advice.

“Follow me for more” asks someone to commit to an ongoing relationship at the exact moment they have seen one post from you. The people most likely to follow are the ones who already decided to, and they do not need the prompt. The cost is that the line takes the last position in your post, the most valuable real estate for an ending, and spends it on something that was going to happen anyway.

“DM me” is better but still asks for a private, one-to-one commitment from someone who has not spoken to you. It works when it is attached to something concrete and small, and it fails when it is an open invitation. “DM me if you want to chat” produces almost nothing. “DM me the word ‘pricing’ and I will send the one-pager” produces qualified conversations, because the reader knows exactly what they are agreeing to and it takes eight seconds.

“Connect with me” has a specific problem: a connection request from a stranger who has seen one post is the weakest connection in your network, and a feed full of them makes your future posts land on people who never really opted in. Growth for its own sake makes your engagement rate worse, since rate is measured against followers. Our engagement figures are all expressed per 1,000 followers for exactly that reason.

A better default: end with something that earns the follow rather than asking for it. A reader who finds the ending useful will look at your profile without being told to, and that is the traffic that converts. If you want the profile to hold up when they do, see how to set up the profile.

How do you test your LinkedIn call to action?

The ending is one of the easier things to test, because unlike the hook it does not change what the post is about.

1. Tag every post with its ending type. Question, soft CTA, direct ask, none. One word in your calendar.
2. Wait for at least eight posts per type. Comment counts are noisy and one post that took off will otherwise decide the answer for you.
3. Compare median comments, not total. Same reason.
4. Read the comments, not just the count.Twenty comments saying “Great post” are worth less than four that describe a real situation, and only one of those two outcomes leads anywhere.
5. Check what happened after. Track whether any comment turned into a conversation, a connection, or a client. That is the number that actually matters and almost nobody records it.

For context on what to expect, our whole-cohort median is 12 comments per post and the 90th percentile engagement rate is 5.95 per 1,000 followers against a median of 0.40. Those come from established creators with 1,000 or more followers, so use them as a direction rather than a target. The full percentile breakdown is in the engagement study.

The short version of the LinkedIn call to action: ask one specific thing that the reader can answer in a sentence, or ask nothing at all. Skip the generic sign-offs, keep the link out of the last line, and remember what the numbers say about where the gap between good and weak posts actually sits. Reactions differ by 3x. Comments differ by 14x. The ending is where you decide which of those you are playing for.