Answer what they wrote, do not mention where you found it, and quote only words they actually used. Those three rules fix most first messages. Below are eighteen rewrites, grouped by the kind of comment you are replying to, with what each original gets wrong.
Every example here is invented. These are not customer messages, not our own outbound, and there are no reply rates attached to any of them, because we have not measured any in a way we could reproduce. Disclosure: Signl is ours, and it drafts openers of this shape; the three code-level rules named below were checked on 24 October 2026.
Key facts
- The source decides what you may say. The comment decides what you should say.
- A quoted fragment must appear verbatim in what they wrote, or it should not be quoted.
- Three sentences is the working length: proof you read it, something new, a small ask.
- Some comments are a reason not to write at all, and recognising those is half the skill.
- A public reply under the comment is usually a better first move than a message.
The three rules
1. Never reveal the source, except on your own post. On your own post, "thanks for the comment on X" is honest and fine. On a creator's post you may mention the topic, not the author. On a competitor's post you say neither. Signl enforces this as a hard guard: a drafted opener on a competitor source that names the post's author or refers to their comment or post is rejected, and a plain line is used instead. Any opener containing a link is rejected too.
2. The quote has to be real. If your line answers a fragment, that fragment must appear word for word in their comment. Signl checks this and falls back to a template when it fails, because a cold message containing a sentence the person never wrote is worse than a cold message with no personalisation at all.
3. Answer, do not pitch. The comment is a statement about their situation. A reply that engages with the situation earns a response. A reply that treats the comment as a trigger for a product paragraph does not.
Rewrites: a stated problem
The most common useful comment, and the easiest to get right.
1. A number in the comment
They wrote: "We cut month end close to four days but reconciliation is still manual."
Before: "Saw your comment on Acme's post about close times. We help finance teams automate reconciliation. Open to a quick chat?"
After: "Four days with reconciliation still manual is the gap I hear about most. In most teams it is two or three specific reconciliations doing the damage rather than the whole process. Worth me naming which ones usually are?"
What changed: no source named, the number answered directly, and the ask is for permission to be useful rather than for a meeting.
2. A vague problem
They wrote: "Our reporting is a mess."
Before: "I noticed you mentioned reporting challenges. Many companies in your industry struggle with this. Would you be open to learning how we solve it?"
After: "Mess usually means one of two things: nobody trusts the numbers, or everyone trusts them and they take a week to produce. Which one is yours? The fixes are completely different."
What changed: the vague comment gets a question that qualifies and demonstrates knowledge at the same time. No claim about their industry that you cannot support.
3. A problem they have already half-solved
They wrote: "We built an internal tool for this, works ok but nobody maintains it."
Before: "Building in-house often ends up costing more than buying. Happy to share a comparison."
After: "The maintenance cliff is the real cost of internal tools, and it usually arrives when the person who built it changes team. Did yours end up with one owner, or is it genuinely nobody's?"
What changed: the before argues with their decision in sentence one, which ends the conversation. The after agrees with the part they already said and asks about the part that decides whether they will do anything.
Rewrites: dissatisfaction with an incumbent
The highest-value comments and the ones most often ruined by mentioning the competitor.
4. A named complaint on the competitor's own post
They wrote: "Been waiting three weeks on a support ticket for this."
Before: "Saw your comment on Acme's post. Their support is famously bad. We answer in under an hour."
After: "Three weeks on a ticket usually means it was escalated and then dropped between two teams. If it is still open, the thing that unsticks it is asking for a named owner rather than an update."
What changed: the before names the source and attacks the competitor, which makes you both a watcher and a sniper. The after helps with the actual problem and says nothing about any product. If they reply, you have earned the next message.
5. Someone who has decided to leave
They wrote: "We're moving off it next quarter, just need to pick a replacement."
Before: "Great timing! We'd love to be considered. Can I send over a demo link?"
After: "The replacement choice usually comes down to whether you need the migration done before the renewal date or can run both for a month. Which constraint are you under? It rules out half the options either way."
What changed: the before asks them to do work. The after proves you have seen this decision before, and the question it asks is one they are already arguing about internally.
6. Dissatisfied but staying
They wrote: "It's not great but switching would be a nightmare."
Before: "Switching is easier than you think! We migrate you for free."
After: "Switching cost is the right reason to stay, and it is usually dominated by one thing rather than the whole project, often historical data or an integration nobody wants to rebuild. Which one is the blocker for you?"
What changed: the before contradicts their stated reasoning with a claim they have no reason to believe. The after takes the objection seriously, which is the only way it ever gets dismantled.
Rewrites: a question
Someone asking a question in public has given you an invitation, so take it literally.
7. A direct question you can answer
They wrote: "Does anyone know if this works with Postgres?"
Before: "Hi! Yes we support Postgres. Here's a link to book time with our team."
After: "Yes, through logical replication, though the catch is that it needs a replication slot and some managed hosts cap how many you get. Worth checking your provider's limit before you commit to the approach."
What changed: the answer is actually useful and includes the caveat, which is what makes it credible. This one belongs as a public reply, not a message.
8. A question about the category, not a product
They wrote: "Is anyone actually getting value out of these tools or is it all hype?"
Before: "Great question! Many of our customers see significant ROI. Want to see a case study?"
After: "The split we see is whether the team had a defined ICP before they bought. With one, the tool filters. Without one, it just produces more rows faster, which feels like hype because it is."
What changed: the after takes a position, which is the only kind of answer a sceptical question respects. It also happens to qualify the reader.
9. A question you cannot answer well
They wrote: "What's the best approach for a 4000-person org with three ERPs?"
Before: "We work with large enterprises all the time. Let's set up a call."
After: Say nothing, or reply publicly that you do not know and name someone who would. Three ERPs at that headcount is a consulting problem, and a message pretending otherwise costs you the credibility you would need later.
What changed: recognising a comment you should not act on is a skill, and it is the one this article's sixth section is about.
Rewrites: a disagreement or a strong opinion
10. They disagree with the post
They wrote: "Disagree, this is just rebranded what we've done for years."
Before: "Interesting perspective! Many people think that until they see what's changed. Open to a chat?"
After: "The part that is genuinely rebranded is most of it. The part that is not is that the tooling now runs without a person in the loop, which changes who can do it rather than what gets done."
What changed: the after concedes the true half, which is the only opening move that works on someone who just called something hype.
11. A strong opinion you share
They wrote: "Per-seat pricing makes no sense when the user is a piece of software."
Before: "Totally agree! Our pricing is usage-based. Want to hear more?"
After: "Agreed, and the version of this nobody prices for yet is an agent that runs a query a thousand times because it is thorough. Thoroughness becomes the line item. Have you hit that yet?"
What changed: the after adds the next thought rather than agreeing and pivoting. We wrote a whole position on this; linking it would be the obvious follow-up once they reply, not now.
12. A strong opinion you do not share
They wrote: "Outbound is dead, everything is inbound now."
Before: "Respectfully disagree, outbound is alive and well with the right approach!"
After: Skip it, unless you have something specific. Arguing with someone's public thesis in their DMs is a poor first impression, and the same person will be reachable on a comment about a problem next month.
What changed: not every strong signal is a strong opportunity.
Rewrites: thin engagement
Short comments and reactions. The temptation is to use the same template; the right move is to use the person's role instead.
13. A two-word comment
They wrote: "This, exactly."
Before: "Thanks for engaging with my post about reconciliation! Would you like to see how we help?"
After, if it was your own post: "Thanks for that. Curious which part landed, the manual reconciliation bit or the close timeline? I am writing something longer on whichever people care about."
After, if it was anyone else's post: Do not write. There is nothing to answer. If their role fits your ICP, use a role-based opener that makes no reference to the comment at all, and accept that it is a cold message.
What changed: the honest acknowledgement that a thin signal gives you nothing to personalise with. In Signl a reaction has to clear a default confidence of 60 before an email is bought, against 40 for a comment, which is the same judgement expressed as a threshold.
14. A reaction and nothing else
They wrote: nothing. They tapped like.
Before: "Saw you liked my post on close times, thought you might want this guide."
After, on your own post: "You liked the close times post, so this may be useful: the two reconciliations that cause most of it are bank and intercompany. No ask, just thought it was relevant."
After, on a stranger's post: Nothing. A like on someone else's post is not a reason to contact a stranger.
15. A repost with a short note
They wrote: reposted with "Important for anyone in finance ops."
Before: "Thanks for sharing our content!"
After: "You added that it matters for finance ops specifically, which is the framing most people miss when they share this. Is that where you sit, or are you selling into it?"
What changed: the note they added is the only original material available, so it is the only thing worth quoting, and the question resolves the pronoun problem before you waste a second message.
Rewrites: the ones not to send
Three comments that look like leads and are not. Recognising them is worth more than any template.
16. The agency or consultant
They wrote: "My clients deal with this constantly."
Do not send: a prospect message. They are not buying for themselves.
Send instead: "Sounds like you see this across a few clients. Are you solving it the same way each time, or does it depend on their stack? I ask because the pattern across clients is more interesting than any single one."
Why: they may be a partner or a reseller, which is a different and sometimes better conversation. "My clients" is the fastest disqualifier in this channel.
17. The job seeker
They wrote: "Great insight, this is exactly what I worked on in my last role."
Do not send: anything. Past tense plus an open-to-work headline is a person looking for a job, and a sales message is at best irrelevant and at worst unkind.
Why: in our experience the more popular the post, the higher the share of these. Signl's free rules set aside headlines saying open to work or actively looking before any model or lookup runs, which means they never consume a verified email either.
18. The person who just bought
They wrote: "We rolled this out last month, happy so far."
Do not send: a switching pitch. Nobody wants to be told their decision last month was wrong.
Send instead: nothing now. Note the date, and be the person who asks how it went in nine months, before the renewal.
Why: this is the single clearest case where the strongest-looking engagement is the weakest opportunity, and where patience beats a template.
Channel order
Public reply first, in most cases. Then a message or an email, and only if the public reply either worked or was never appropriate.
A public reply is proportionate, visible to everyone else reading, and requires no contact data. Signl drafts one under 25 words with no pitch, no link and no product name, and only where a comment exists, never for a reaction. We think it is the better first move more often than teams expect.
Email suits the stated-problem comments, where there is no question to answer and no thread to join. LinkedIn messaging from your own account inside Signl is listed as coming soon on our pricing page; today that sending happens in your own HeyReach or Expandi account, and Signl routes people there.
What never works
- A line that begins "Saw your comment on".
- A quote the person did not write.
- A first message containing a link. Signl rejects drafted openers with links for this reason.
- Four sentences. Three is the ceiling on a first message.
- The same opener on a comment and on a reaction, which tells the reader exactly how it was produced.
- Mentioning the competitor whose post you found them under, ever.
The sourcing side of this, including which comment sections are worth reading in the first place, is finding leads in competitors' comments and the wider LinkedIn social listening guide.
This article is about the line. If the message is going to an inbox rather than staying on LinkedIn, the channel switch changes what you may say and what the subject line has to do: turning a comment into a cold email.
