A security team posted on Reddit that four people cover about 60 EKS clusters, sit on 1,200 critical findings and face a renewal in six weeks. That post names the problem, the competitors and the deadline. Social listening finds it. Answer it in public, then DM the author, and you skip the cold sequence entirely.

TLDR

Buyers post their problem in public with a job context attached, and you can find those posts for free.
Score every post on relevancy, authority and velocity before you spend time on a reply.
Reply in public for the silent readers, then DM the poster with a message that quotes the post.
Map the handle to a company and check two lists before anyone messages.
Keep a person in the loop on every reply, or the program gets banned.

A Scored Thread Replaces a Cold List

Take one goal, a meeting with a buyer for a cloud security product. Run it two ways.

The cold route starts with a list of 40 contacts. You pay for a rep's time, a sequencing tool and a data vendor, and you interrupt strangers who never asked.

The listening route starts with one thread. A buyer wrote down the problem, so your reply and your DM both answer a question that exists. Marketers call this demand capture, and it costs minutes of careful writing.

Rule of thumb. If you must convince the reader that they have a problem, you are on the cold route.

Step 1. Listen Where Buyers Post

A listening tool watches Reddit, X and LinkedIn for posts that match your product, then ranks them. I use SensorHub for AccuKnox, where it scans about 1,700 conversations a day across those three platforms. Most of them earn no reply.

The SensorHub card on the tools page lists it as a watcher of Reddit, X and LinkedIn for buying conversations.

The feed below came from one project in October 2026. The score runs from 0 to 1 and measures the match to the product.

A listening feed ranks public threads, so you read the top one first.

Four threads scored 0.95 or higher. Each one names a competitor or a deadline. The fifth scored 0.55 and names neither.

One builder describes public complaints as signals you can monitor.

Step 2. Score Each Post Before You Reply

A high match does not make a post worth your time. Score three dimensions, and each one filters out a different kind of waste.

Relevancy. Does the post name your category or a competitor? A generic question scores low.
Authority. Does the poster hold budget? A team size, a renewal date or a title says yes.
Velocity. How many people are in the thread? A busy thread reaches a large silent audience.

Score each post on relevancy, authority and velocity, and reply only to the one that passes.

The left thread passes on all three. The right thread fails on all three, because its author builds GRC software and asks questions for research. SensorHub gives the relevancy score and the comment count, and I judge authority by reading the post.

Velocity is the dimension teams underrate. Jakob Nielsen's participation inequality article, published in 2006, found that about 90% of users in a community only read. A thread with 20 comments reaches a far larger audience than the people who typed.

Tip. Scoring is cheap when a model does it. In What Is Jev I describe a model that returns a score instead of prose, which fits this job. The Appendix has a prompt that works with any model.

Step 3. Reply With Help, Then DM the Poster

A passing thread gets two messages that answer the same post. Neither one pitches.

The public reply serves the silent readers, and the DM serves the poster.

Public reply. Answer the exact question with a direct comparison, a diagram or a snippet. The silent readers see it too.
Private DM. Quote the post in the first line. A message that names the poster's own words gets answered.

Match the reply to the post type. A competitor complaint gets a comparison that targets the gap the poster named. A technical wall gets an architecture diagram or a working fix. A category question gets an honest map of the space, because that poster is still forming a view.

Another founder describes DMing people who had already complained in public.

Step 4. Turn the Handle Into a Pipeline Row

A Reddit username tells you nothing until you map it to a person. Social enrichment tools do that, because people reuse handles and link other profiles.

Two list checks stop you from pitching the wrong account.

Run both checks before anyone sends a message. The target list tells you whether the account can ever close. The customer list stops you from cold pitching a company that already pays you.

Log each thread in a CRM row with the link, the score and an owner. If you want the same job done from a chosen list of accounts, read the LinkedIn comment pipeline.

My Signals Proposal skill turns a feed of public buyer conversations into ranked actions, content ideas and keywords.

The Signals Proposal page shows ranked actions, each with an owner and a reason.

To see the whole workflow in one tool, read the 3-step SensorHub workflow for Reddit and LinkedIn.

Step 5. Keep Replies Human to Stay Unbanned

Run this at scale with no guardrails and a subreddit bans you. Three rules prevent it.

A person reads every reply before it posts. A model can draft the first pass, and a person checks that no line reads like marketing copy.
Spread the work across real people with real expertise. One brand account that answers every thread triggers a spam flag.
Check every handle against the customer list first.

Watch out. Never auto-post. Even the comment tools I tested for LinkedIn needed a review step, which I cover in I Almost Spent $250 on LinkedIn Auto-Comments.

A reply on a public thread also outlasts a paid campaign. It stays online and keeps answering the same question for the next buyer who searches it.

Questions Marketers Ask About Social Listening

Do I need a paid tool to start? No. Search your category and competitor names on Reddit, X and LinkedIn each morning, then score what you find with the three dimensions. A tool saves time once the volume grows.

Will a reply show up in AI answers? Possibly, though I have no citation count. Reuters reported in 2024, citing sources, that Google pays Reddit about $60 million a year to license its content, and neither company confirmed the figure at the time (Search Engine Land summary). OpenAI announced a Reddit partnership in May 2024. Write replies worth citing, and treat any citation as a bonus.

OpenAI announced the Reddit partnership on its own account.

Is the DM too pushy? The DM quotes a post the person wrote in public and offers help with that problem. Skip any thread where the poster is a vendor or has no buying context.

What does a reply cost? One careful answer, which takes minutes. The cost that matters is the review step, and that one does not shrink as volume grows.

The Appendix Holds the Score Prompt and the DM Template

This scoring prompt works with any model. Paste the post below it.

Score this public post for a buyer of [your product category].
Return three scores from 0 to 1 and one line of reason for each.
1. Relevancy: does it name the category or a competitor?
2. Authority: does it show budget, team size, a deadline or a title?
3. Velocity: how many comments does the thread have? (I will paste the count.)
Then answer one question: reply and DM, or skip?

A DM opening that quotes the post:

I saw your post about [the exact problem in their words].
We worked through that with a team of a similar size.
Want the steps we used?

How to Do Social Listening for Leads in 5…