Comments on my best post
260
Sold what I sell
121
Had a buyer's job title
38

Your next 500 buyersare in someone else's comments

People who comment on a post about your buyer's problem are telling you they have that problem. This is how to collect them from the posts that already exist, cut out everyone who can't buy, and start a conversation with the rest.

Free · 1 Claude skill · 2 ways to pull comments · Connection notes · My own run

01 · Read this first

A comment is a raised hand

By the end of this you'll have a sorted list of people who commented on posts about your buyer's problem, with a connection note ready for each buyer.

Cold lists from a database tell you someone has the right job title. A comment tells you they cared about the topic this month, and you can open with what they said instead of a template.

You'll need Claude (the desktop app if you want the free browser option), a short buyer file you fill in once, and either an Apify account or about an hour of Claude clicking through posts for you.

What's inside

  1. My own run, with the numbers, so you can see why the sorting matters
  2. How to pick the posts whose commenters are your buyers, not your peers
  3. Two ways to pull the comments: a scraper that costs about $2 per 1,000 comments, or Claude's browser for free
  4. The skill that scores every commenter 0 to 100 and sorts them into Buyers, Maybes, Peers and Skips
  5. The connection notes it writes from each person's comment
  6. The sending plan, so your account stays safe

02 · My own run

260 comments, 38 possible buyers

I tested this on my own best post first: a lead magnet about opening lines that got 732 comments with replies, and 260 top-level comments from other people.

Who left the 260 comments
  • Sell what I sell (content, LinkedIn, design, video) · 121
  • Other roles, mostly freelancers and employees · 94
  • A buyer's job title outside my space · 38
  • Students, job seekers, empty headlines · 7

218 of the 260 typed the keyword and nothing else. If I'd sent a connection request to everyone, almost half would have gone to people who do my job.

The lesson is about the post more than the sorting. A post about writing better hooks pulls writers. A post about your buyer's problem pulls buyers. So step one is picking the right posts.

03 · Pick the posts

The post decides who's in the comments

Pick 10 to 20 posts. Ten good ones beat a hundred random ones.

Pick posts thatSkip posts that
Come from creators your buyers followCome from creators other marketers follow
Talk about your buyer's problem in their wordsTalk about LinkedIn, content or AI tools in general
Have 50+ comments from the last 30 daysAre months old, so nobody remembers commenting
Ask a question or offer something your buyer wantsAre giveaways anyone would grab

To find the creators, look at who your best clients follow and comment on, or ask Claude to search LinkedIn posts for your buyer's problem and list the authors who keep coming up.

Prompt to shortlist the posts
Use the buyer-list skill. Here are the creators my buyers follow:
{paste profile links}

Open each profile's recent activity, list their posts from the last 30 days with 50+ comments, and pick the 15 whose topic is closest to the problems in my buyer file. Skip anything about LinkedIn growth, writing or AI tools unless that's what I sell.

04 · Pull the comments

Two ways to get every commenter

Use the scraper when you have more than a few posts. Use Claude's browser when you want it free and don't mind waiting.

Option A: the scraper (fast, about $2 per 1,000 comments)

  1. Make a free account at apify.com. New accounts come with a small monthly credit, which covers a few thousand comments.
  2. Open the actor harvestapi/linkedin-post-comments. It doesn't need your LinkedIn login.
  3. Paste the post links into Posts, set the maximum comments per post, leave Scrape replies off and run it.
  4. Export the dataset as CSV and give it to Claude. Or connect Apify to Claude and ask Claude to run it for you.
Input for the scraper
{
  "posts": [
    "https://www.linkedin.com/posts/...",
    "https://www.linkedin.com/posts/..."
  ],
  "maxItems": 500,
  "scrapeReplies": false,
  "profileScraperMode": "short"
}

My 260 comments came back in 51 seconds. Each row has the person's name, headline (in a column called position), profile link and comment.

Option B: Claude's browser (free, slower)

The Claude desktop app has its own browser. Claude opens the post, loads the comments and reads them for you. It runs as you, signed in to your own LinkedIn, so keep it slow and keep it to a few posts a day.

Prompt for the browser
Use the buyer-list skill. Open this post in your browser:
{post link}

Click "Load more comments" until no more load, waiting a few seconds between clicks. Stop at 300 comments. For every top-level comment, write down the name, headline, profile link and comment text. Then do the same for the next post:
{post link}

When you're done, sort everyone with the skill.

05 · Sort them

Install it and sort the list

The skill reads your buyer file, scores every commenter from their headline and comment, and writes the notes.

  1. Download buyer-list.zip below and don't unzip it.
  2. In Claude, go to Customize > Skills, click +, then Upload a skill, and pick the zip.
  3. Start a chat and say "use the buyer-list skill". The first time, it asks the questions in buyer.md and saves your answers.
buyer-list.zipSKILL.md and the buyer file templateDownload

If your plan doesn't have skills, make a Claude Project, paste the SKILL.md text into the project instructions and add buyer.md to the project. It works the same way.

How a commenter's score is built (out of 100)
  • Buyer job title+35
  • Right company type+20
  • Says more than the keyword+15
  • Names your buyer's problem+15
  • Names a company+10
  • On 2+ of your posts+5
Takes points offPoints
Spam or an ad for their own thing-50
Sells what you sell, or on your "never buys" list-40
Student, job seeker or empty headline-25
TierRuleWhat you do
Buyer60 and upConnection request with a note
Maybe35 to 59Request after the Buyers, note optional
PeerSells what you sell, whatever the scoreLeave them
SkipUnder 35Leave them

You get buyer-list.csv with name, headline, link, the post they commented on, their comment, score, tier, the reason and the note. Read the Buyers before you send anything, since the skill only sees headlines. If you move someone, tell Claude why and it keeps that rule for the rest of the chat.

06 · Send the requests

15 to 20 a day, Buyers first

A note that mentions what they said gets accepted far more often than a blank request, and it gives you a reason to talk later.

Two notes the skill writes (made-up people)
Hey Sarah, saw your comment on Tom's post about hiring a first salesperson. Doing it before you have a playbook is brave. Would be good to be connected.

Mark, your point about cash flow in month 3 on Priya's post stuck with me. I work on the same problem with agency owners. Happy to connect?
  • Send 15 to 20 requests a day, Monday to Friday. LinkedIn caps invitations per week, so check where yours sits before you go faster.
  • Start with the Buyers who commented most recently.
  • Send them by hand, or load the list into the tool you already use for connection requests.
  • When someone accepts, don't pitch. Reply to what they commented and ask about their situation.
  • Run it again next week with new posts. Ask Claude to skip anyone already on the list.