ICP Discovery

Find and Rank Your Best Prospects from LinkedIn Connections

What this achieves

You want a ranked, scored list of your LinkedIn connections that match your ideal customer profile - not just a headline match, but actual buying signals from what people are posting and doing right now. Success is a table of qualified prospects ranked by how likely they are to buy, with specific notes on why each one scored where they did.

Define your ICP buckets

Pick 2-4 customer types you're hunting for. For each one, write a one-sentence description of who belongs AND who doesn't. Example:

Solo Operator: runs their own shop - founder, freelancer, independent consultant. Not a VC, not an employee at a big company, not someone selling hardware.

Boutique Agency: small shop serving B2B clients (design, dev, growth, GTM). Not a corporate governance firm or a quant fund.

IT Consultant / MSP: independent IT consultant, RevOps person, managed services. Not a Solutions Architect employed at a vendor.

Phase 1: Scan your connections

Go to your LinkedIn connections page. Scroll deep - you'll need to scan roughly 3x the number of ICP matches you want. If you want 100 good prospects, expect to scan about 300 connections. Capture name, headline, and profile link for each one.

Phase 2: Sort them into buckets

Match each connection to an ICP bucket based on their headline. Use keyword matching, but keep it conservative - you'll catch more false positives than false negatives. "Founder" matches VCs and service providers just as easily as solo founders. "Agency" catches corporate governance firms. Err on the side of narrower matches; you can add missed ICPs later.

Give each match a preliminary score based on their headline alone, from 1 to 10. Be skeptical. Most headline scores will drop after you look at actual activity.

Phase 3: Enrich with real buying signals

This is where the work pays off. For each prospect, visit their LinkedIn recent activity page. Read what they're actually posting and commenting about right now. The headline tells you who they are. Recent activity tells you whether they're ready to buy.

What counts as a buying signal, ranked strongest to weakest:

1. Explicit pain. They're posting about something broken: "My CRM migration wrecked our pipeline," "proposals take 3 days because nobody owns the template," "reps are back to spreadsheets."

2. Tool complaints or switching. They're vocal about leaving one tool for another, evaluating options, or frustrated with their current stack.

3. Active hiring. Especially sales, GTM, or growth roles. Signals scaling - scaling means budget.

4. AI or automation as a user. They're posting about workflows they built, agents they deployed, or Claude/Cursor tricks they discovered. They're a practitioner, not someone selling AI services.

5. Building and launching. Y Combinator, shipping product, seeking a co-founder, running events.

6. Explicit asks. "My DMs are open," "looking for recommendations," "anyone using X?"

What counts against them - downgrade the score:

- Every post has a "Book a demo" or "Schedule a call" CTA. They're selling, not buying.
- All content promotes their own company or product. Service provider.
- No posts in six months or more. Dormant profile. Cap score at 5.
- Only reposts, zero original thoughts. Curator, not buyer.
- Wrong ICP entirely - a VC posting about portfolio companies, a corporate compliance person. Score 2-4.

Phase 4: Re-rank and review

Once every prospect has been enriched, re-sort the full list by final score. You'll typically see:

Score 9-10: Active builders with explicit pain. Your best outreach targets - probably 10-20% of the list.

Score 7-8: Good fit, relevant activity, but no urgent need expressed. Nurture over time.

Score 5-6: Right headline, wrong reality - dormant, or a service provider with adjacent content. Low priority.

Score 2-4: Wrong ICP. Headline misled the classifier. Remove or re-bucket.

Also flag profiles that landed in the wrong bucket. A solo founder posting agency case studies probably belongs in Boutique Agency. A VC sharing portfolio news doesn't belong anywhere.

The most important distinction: buyer vs. provider

About 30% of your headline matches will turn out to be people who sell to the same customers you sell to - Clay partners, HubSpot consultants, fractional CMOs, GTM engineers. They use the right vocabulary. They post about the right problems. But they're looking for clients, not tools. Enrichment must explicitly flag "service provider, not buyer" for each prospect where this applies. These people can be partners or referral sources, but they're not outreach targets.

Scoring reference card

10 - Building something. In pain. Hiring or switching tools. Clear buying intent right now.

9 - Strong fit, active, engaged. Likely in market within 90 days.

8 - Active and relevant, but no urgent need visible yet.

7 - Good fit, some activity, no pain signals.

6 - Right headline but dormant profile, or service provider with adjacent content.

5 - Borderline fit, or completely dormant. Headline-only match.

2-4 - Wrong ICP. Remove from outreach list.

The final deliverable

A ranked table of your ICP connections with: name, headline, bucket, score, and a one-sentence note on why they scored that way. The note is the most valuable column - it's what turns a number into a conversation starter.