Watch four kinds of LinkedIn post: your own, your competitors', those by people your buyers follow, and topic searches. Collect the commenters, drop job seekers, vendors and empty praise, and check the rest against your ICP before buying an email. Then open with a line that answers what they wrote, and never say where you found them.
Disclosure: Signl is ours. The seven steps work with any tooling, including a spreadsheet and patience. Where Signl does a step, the exact setting is named, verified against our code on 4 October 2026.
Key facts
- A comment gives you words to answer. A like gives you a headline and nothing else.
- Signl watches five kinds of LinkedIn source: a profile, one post, a topic search, mentions of a company, and a company's staff posts.
- LinkedIn links containing
-share-cannot be read for comments. Watch the author's profile instead. - Qualifying before enrichment is the cost control. In Signl, a filtered person never gets an email lookup, so the plan's monthly verified email allowance goes only to people who passed your ICP check.
- An opener for someone found on a competitor's post should never name the competitor or the post.
Why is a comment warmer than a like?
We think a comment is the warmest public signal LinkedIn offers, because it is the only one you can reply to. "We cut month-end close to four days but reconciliation is still manual" tells you a problem, a stage and a vocabulary. A like tells you the person scrolled past and tapped.
Our own pipeline is built on that view. When one person both commented on and reacted to a post, Signl keeps the comment. A reaction has to look like a buyer from the headline alone, and has to clear a default confidence of 60 before any email is bought, against 40 for a comment. Likes are still worth collecting from your own posts, where the relationship already exists. From strangers' posts, comments carry the weight.
Step 1: Decide whose posts to watch
Four relationships, and each one changes what you can say later.
| Relationship | Example source | What you get | What the opener may say |
|---|---|---|---|
| Own | Your profile, your company's posts | People already engaging with you | That they commented on your post |
| Competitor | A competitor's founder, or posts mentioning the competitor | People with the problem you solve, often mid-evaluation | Only the substance of what they wrote |
| Influencer | Someone your buyers follow | Buyers thinking out loud in a trusted space | The topic, not the author's name |
| Other | A topic search such as "month end close" | Strangers using your buyers' words | The topic, not the author's name |
A topic search finds posts, not just comments, so the people who wrote those posts count as leads too. Signl also searches for posts mentioning a company, and for posts by a company's staff. A mentions search is labelled competitor by default.
Set the relationship correctly when you add a source. In Signl it cannot be edited afterwards, because it decides what openers may say. You remove the source and add it again.
If you do not know whose posts your buyers read, start with your own and one competitor. After the first collections, Signl suggests authors that two or more of your ICP people engaged with in the last 30 days. The rules are in Sources Signl finds for you.
Step 2: Paste a link that can actually be read
LinkedIn's share button produces several URL shapes. Some contain -activity- and some contain -share-. Comments on a -share- link cannot be read, so Signl refuses that link and asks for the author's profile URL instead. We chose a refusal over a source that silently returns nothing.
A profile is usually the better source anyway. Signl checks a watched profile for new posts about once a day, so next week's post is covered without you pasting it. Use a single post when one specific post matters, and pick scrape once if you only want its commenters today rather than as they keep arriving. If the post is your own lead magnet, where everyone who comments was promised something, delivery works differently: see LinkedIn lead magnet posts.
Step 3: Choose comments, likes or both
Each source collects comments, likes or both. Signl is priced by the number of sources you watch, not per person, and there are no credits: everyone a source finds, their qualification and their verified emails are included within a monthly fair use.
| Plan (as of 24 October 2026) | Price | Sources | Checked | People included | Seats |
|---|---|---|---|---|---|
| Self-serve | $99/mo, or $79 billed yearly | 25 | Hourly on up to 15, the rest daily | Every commenter, plus 2,000 likers a month | Unlimited |
| Signal Engine | $2,000/mo | Sized to you | Hourly | The same, run by our team | Unlimited |
A topic search counts as 2 sources. Billed yearly, Self-serve is $79 a month ($950 a year), and more than 25 sources is arranged by hand on a custom plan. There is no free plan: paste a LinkedIn profile on the home page to see who engaged with its recent posts before you buy. The fair use behind the plans is on the fair use page.
Our suggested starting configuration: comments only on competitor, influencer and topic sources, both on your own posts. A viral post from a large influencer is read up to 500 people per check on every plan, and only the people who pass your ICP get an email lookup.
Step 4: Throw out the noise for free
In our experience, most comment sections hold more noise than leads. Strip it with rules that cost nothing before any model or lookup runs. Signl's free rules set aside:
- Comments with no text, or only emoji or punctuation
- Generic praise of six words or fewer ("Great post", "So true", "Congrats")
- The post's own author replying under it
- Headlines that say open to work or actively looking
- Recruiters, unless your ICP sells to recruiters
A classification step then reads each remaining comment against your ICP. Anyone judged a job seeker, someone selling their own service, or noise is never enriched, however confident the model is. Every set-aside person keeps a reason in words, in the Filtered out tab, so you can see a wrong call and correct your ICP.
Step 5: Qualify before you pay for the email
This is the step to get the right way round. Several playbooks ranking for this query on 4 October 2026 enrich every engager first and score afterwards. Done in that order, you paid to look up the job seekers.
Signl runs four phases in a fixed order: discover, classify, enrich, compose. Classification decides whether to look up an email at all, and anyone it filters out never uses your plan's email allowance. A person who passes gets the verified work email lookup, included in your plan within fair use. Why paying for results rather than attempts matters is covered in the bounce tax, and what "verified" should mean in what verified email means.
Your ICP is the filter, so it deserves ten minutes: what you sell, the titles you sell to, company size, countries and your competitors. A known country outside your list caps the match, whatever the model thought.
Step 6: Write a first line from what they actually wrote
A grounded opener picks up a specific fragment of the comment. Signl enforces three rules in code, not just in the prompt:
- The quote must be real. The fragment the opener answers has to appear verbatim in the comment. If it does not, a plain template line replaces it. An invented quote in a cold email is worse than none.
- Where it was said decides what may be said. On your own post, the opener can say so. On anyone else's, it may mention the topic but not name the author.
- Competitor sources stay silent about the source. A drafted opener that names the post's author or mentions their comment or post is rejected, and so is any opener containing a link.
An illustrative example. Someone writes under a competitor's post: "We cut month-end close to four days but reconciliation is still manual." A line that will make them feel watched: "Saw your comment on Acme's post about close." A line that answers them: "Four days with reconciliation still manual is the gap I hear about most."
We think a public reply under the comment, before any email, is often the better first move. Signl drafts one under 25 words with no pitch, no link and no product name. Comment replies are only drafted where a comment exists, never for a like.
Step 7: Send them to the tool you already use
Signl sends people to Instantly, Smartlead, lemlist, HeyReach, Expandi, HubSpot, Clay or any webhook, and can post alerts to Slack. Autopilot is a rule of the form: people at or above this confidence, from these kinds of source, go to this campaign, at most this many a day. It runs after each collection, and never sends the same person to the same campaign twice. An account can have up to 10 rules and 500 autopilot sends a day across them.
A sensible first rule is your own posts only, at a high bar, into a LinkedIn tool such as HeyReach. Add competitor sources once you have read twenty of their openers yourself.
What goes wrong, and how to spot it
Job seekers and vendors fill the thread. In our experience, the more popular the post, the more of both. Read the Filtered out tab after the first collection. If good people sit there, the fix is your ICP, not a lower bar.
Engagement pods. LinkedIn's Professional Community Policies say not to "agree with others ahead of time to like or re-share each other's content" (read 4 October 2026). Where pods operate, their comments look like engagement from the same few names on every post. Signl has no pod detector. The generic-praise rule catches some pod comments. If one author's comment section is the same twenty people every week, drop that source.
Share links that return nothing. Covered in step 2. Watch the profile.
Competitor openers that creep people out. We think the fastest way to burn a warm signal is to tell someone you were watching. Keep the competitor label accurate, since that is what switches the guard on.
Reaching a fair use limit. When the likers limit runs out, likers wait until the 1st of the next month and comments carry on. When the verified email limit runs out, email lookups wait until the 1st. Nothing is charged per person.
Objections. Someone you contact may object. The UK ICO's right to object guidance calls an objection to direct marketing "an absolute right" with "no exemptions" (read 4 October 2026). Signl collects only public engagement and public posts, deletes them 90 days after first collection, and once someone objects, collects nothing of theirs again for any customer. This is not legal advice. The lawful-basis side is in GDPR and B2B prospecting.
What to expect
We do not publish reply rates, because we have not measured them across customers in a way we could reproduce. What you should see after the first collection is concrete. You get a list of people with a one-sentence reason each, a set-aside list with a reason each, and openers you can check against the comment they quote.
If you are moving from Trigify, the same steps apply. Trigify alternative covers what to export before it closes on 22 October 2026.
Four companion articles go deeper than a step each allows: LinkedIn social listening for the whole method, LinkedIn buying signals for telling a strong signal from a weak one, finding leads in competitors' comments for step 1's hardest source, and eighteen opener rewrites for step 6.
