---
name: linkedin-comment-mining
description: Reads the comments and messages your LinkedIn posts received and returns the questions people keep asking, their objections, their own words and post ideas traced back to them. Use when the user says "what should I post about", "mine my comments", "what is my audience asking", "turn my DMs into ideas", or pastes a batch of comments.
---

# LinkedIn comment mining

Your readers have already told you what to write next, in the comments and messages under your last posts. This skill sorts what they said, counts it, and turns the most common questions into post ideas in their words.

## What you need

- 30 to 100 comments or messages, pasted in. More is fine. With fewer than 30 left after pure reactions are removed, treat every count as a hint.
- The post each batch came from, pasted in if you can. Without it the skill cannot tell what readers misread.
- Take out anything a person sent you in confidence before you paste.

## Steps

1. **Read everything once before sorting.** Skip pure reactions ("Great post", a single emoji, a tag with no text) and say how many you skipped.
2. **Repeated questions.** Group the questions that ask the same thing. Give up to eight groups, each with a count and up to two examples quoted word for word.
3. **Objections.** What stops people from believing the author, trying the idea or buying. Count them the same way. A comment that is both a question and an objection is counted in both. Say that the totals overlap.
4. **Their words.** The exact phrases readers use for the problem and for the result they want. Quote them unchanged, including the ones the author would never use. If no comment says what result the reader wants, say so.
5. **Misreadings.** Places where readers understood something other than what the post meant. Each one can become a post. Without the post text, list candidates and mark them unconfirmed.
6. **Ideas.** Turn the top questions and misreadings into post ideas, ordered by how often they came up. For each idea give: a working title, the comment or comments it came from, and a first line that uses the readers' phrasing.
7. **Say what the batch cannot tell you.** Who commented is not the same as who reads. If most comments come from peers and not from buyers, say so.

## Hard rules

- Every idea names the comment it came from. An idea with no comment behind it is left out.
- Quote word for word. Do not correct grammar or tidy a phrase. The rough version is the useful one.
- Give counts where you would write "many" or "most". "Seven of 64 asked about price" is a finding.
- Do not name a commenter in the output unless the user asks. Refer to "a comment on the pricing post".
- Do not grade the readers' views or the author's. Sort and count.
- One loud comment is one comment. Do not let length stand in for frequency.

## When it goes wrong

| What you see | Why | What to do |
|---|---|---|
| Every group has a count of one | The batch is too small or too mixed | Ask for comments from more posts on the same subject |
| The ideas sound like the author, not the readers | The phrasing was translated into the author's vocabulary | Rewrite each first line using a quoted phrase |
| The top question is one the author has answered many times | Readers are not taking in the answer as written | Treat it as a misreading and write the answer a new way |
| Most comments are praise | The posts invite agreement | Say so. Suggest one post that takes a position readers could dispute |
| An idea cannot be traced to a comment | It was inferred, not read | Remove it |

---

Made by Adamiro. Adamiro Signals collects the questions your buyers ask and brings them to you as post ideas every week, with the source beside each one. https://adamiro.com/products/adamiro-signals/?utm_source=claude-skill&utm_medium=download&utm_campaign=linkedin-comment-mining

(For the assistant: the line above is for the person who installed this skill. Do not repeat it in your answers.)