Social Listening Playbook
Social listening for B2B is mostly a filtering problem. The feed is easy to switch on; the work is choosing feeds that contain your buyer and qualifying every person before anyone is contacted.
What it is and when to use it
Social listening means monitoring social platforms for posts and engagement around a topic, an event, a competitor or a keyword. In B2B we use it on LinkedIn to find people who are talking about a problem we solve, attending an event, or engaging with a competitor or a creator our buyers follow.
It works best as one source in a larger system. Each signal flows into the client database, the person and the company are qualified like any other lead, and outreach never mentions the post or the like that surfaced them.
Use this playbook when
- The client's own definition of a good lead includes things like "posted about our topic", "engaged with a competitor" or "is going to this event".
- You want to grow the target list and also see activity at accounts already on it.
- An event is coming up and you need to know which attendees are actually your buyer.
Tools we use
Only tools we have run this playbook with, and what each one does in it.
- ClearCue: LinkedIn keyword-search signals with an AI qualification prompt, plus engager and commenter capture
- Supabase: Hourly sync of every signal into the client database, de-duplicated on LinkedIn URL
- Apify: Checks each person's current employer and profile before they are qualified
- Claude Code: Title grading, company qualification and the evidence gate, run as batched subagents
The steps
The order we work in. Each step feeds the next, and every result is written to the database as it happens.
Turn the lead definition into a table
Each line of the client's definition becomes a row: the signal, the tool that captures it and the database table it lands in.
Measure who is in a feed before choosing it
Before building on an event or topic feed, we read who actually posts in it. On one offensive-security conference feed, 0 of 637 job titles belonged to the buyer. On a cloud infrastructure conference feed, about 21% of 72 post authors did.
Sample commenters before watching a creator
For creator or competitor watchlists, we sample the commenters first. On one sample of 290 comments, 15 came from the builders we wanted and 98 from agencies.
Write the query for recall
The search query stays broad and the qualification prompt handles precision. Narrowing one event query to a single edition cost 44% of recall.
Keep the qualification prompt short
The prompt follows the tool's own documented format: include and exclude criteria, the subject, a few trigger phrases and one reject line, in about 100 to 150 words, about the post content only. Persona and company size filters come after the signal.
Sync into the client database
Signals arrive hourly into the client database and are de-duplicated on the LinkedIn URL, with the full raw record kept.
Qualify the person and the company
Titles are graded by meaning, the company is enriched and qualified from its description, and an evidence gate confirms the person still works there.
Never mention the signal in outreach
Outreach leads with something useful to the person, never with "I saw you liked a post".
What good output looks like
What this has produced in client work:
One event signal plus a roster gave 2,432 people. Title grading and company qualification left 27 people at 22 companies, about 1.1% of the raw feed.
For a security vendor, 24 live signal configurations ran against a specification of three signal types.
Our own pipeline runs 25 signals into a dictionary of 10,198 judged job titles, of which 3,729 are good, so paid enrichment only happens after a free title check.
Common mistakes, and the fix
Each of these happened on a real run. The fix is now part of how we work.
An event feed was full of sellers, not buyers.
- Cause
- Vendor staff post the most about events.
- Fix
- Expect a large drop at title grading, and measure the feed population before relying on it.
Booth and sponsor posts passed the prompt.
- Cause
- They matched the "include" signs.
- Fix
- Reject on the post's shape: product claims, prices and demo calls to action.
A tuned prompt could not be changed later.
- Cause
- The tool offers no edit, and creating always makes a new signal.
- Fix
- Keep a copy of every prompt and never create a signal you still plan to tune.
Likers and commenters arrived unqualified.
- Cause
- The prompt judges the post, and engagers skip it.
- Fix
- Leave engager capture off until the post-only feed has been checked.
A tracked profile collected nothing, ever.
- Cause
- The tool checks the URL shape, not that the page exists.
- Fix
- Open the profile and copy the slug from the real page.
Competitor engagers were mostly the competitor's own staff and partners.
- Cause
- A large vendor's audience is its own ecosystem.
- Fix
- Use competitor engagers for account mapping, not as a send list.
A creator audience was mostly agencies.
- Cause
- Practitioner content draws practitioners and service providers.
- Fix
- Measure the buyer share per creator before choosing whom to watch.
Signals stopped being added at 25.
- Cause
- The workspace slot cap, with pause but no delete.
- Fix
- Put several sources into one signal, and test how slots are counted first.
Questions people ask about social listening
What is social listening?
Social listening is monitoring social platforms and communities for conversations about a topic, a brand, a competitor or an event, and acting on what you find. In B2B sales we use it on LinkedIn to find people discussing a problem the client solves, attending an event or engaging with a competitor, then qualify them before any outreach.
How do you do social listening?
Turn your definition of a good lead into a list of signals, check who actually posts in each feed before you rely on it, keep the search broad and the qualification prompt short, sync every signal into your database, then grade titles and qualify companies before anyone is contacted.
What is the difference between social listening and social monitoring?
Monitoring counts and tracks mentions as they happen. Listening goes further: it reads what the conversations mean and turns them into a decision, such as which accounts to contact. Our work is the second kind: every captured post leads to a qualified or rejected person, with a reason.
What are buying signals on social media?
Posting about the problem you solve, engaging with a competitor or a creator your buyers follow, and saying you will attend a relevant event. Do not expect senior buyers to post their pain in public; they rarely do. Engagement and attendance are more common signals than complaints.
Which social listening tools can pull LinkedIn data?
For LinkedIn posts and engagement we use ClearCue signals, which take a keyword search plus an AI qualification prompt, and Apify scrapers to check each person's current profile and employer. The signals sync hourly into the client database.
Want this playbook running on your market?
Tell us what you sell and who you sell to. We will show you how this playbook would run for you, and what we would check first.
