// Playbook
LiveScore LinkedIn Leads by ICP with AI | AtomSell
// When to use
Problem
SDR leads, RevOps and founders who build lists in Sales Navigator.
A Sales Navigator search returns hundreds of profiles and most of them are not a fit. Teams either message everyone and burn their daily sending capacity, or pay to enrich and score every row in a tool like Clay, including leads that will never qualify.
// What you get
Outcome
- Runs the search you already use and saves the results to a data table
- Applies free rules first to drop obvious non-fits
- Scores the remaining leads 1 to 10 against your ICP with a one-line reason
- Writes a connection note only for leads that score 7 or higher
- Keeps rejected leads with their reason, so you can sharpen the ICP
// Results
Proof
// Building blocks
AtomSell Actions used
Search people
Search people on classic LinkedIn or Sales Navigator with every filter, or run any search URL. Ask your agent: "Find Heads of Sales at SaaS companies with 11 to 200 employees in Germany on Sales Navigator."
Look up search filters
Look up ids for locations, industries, titles, companies, schools, functions and skills.
Create a table
Create a data table with typed columns and a key that prevents duplicates. Ask your agent: "Create a table called Event leads."
Write table rows
Add or update up to 1,000 rows at once. Ask your agent: "Save the scores and reasons to the table."
Enrich a profile
Open a full profile: experience, education, skills, languages, certifications, summary, contact details where visible, and how they relate to you. Ask your agent: "Open the full profiles of the top 5 and tell me why each fits."
Enrich a company
Open a company page: description, website, industry, headquarters, founding year, specialties, followers and employees. Ask your agent: "Research these 3 companies and score them against my ICP."
Send a connection request
Send a connection request with a personal note of up to 300 characters. Ask your agent: "Send the notes to the top 10."
// How it works
Flow
// Run it
Implementations
Connect AtomSell to Claude (Settings, Connectors, Add custom connector, https://app.atomsell.ai/mcp), connect your LinkedIn profile in AtomSell, then paste the prompt into a new chat. Start your 3-day free trial at app.atomsell.ai/register.
Prompt
Run this end to end without asking me anything. If you need to choose, pick the fastest option and tell me which. 1) Search Sales Navigator for [TITLES] at [COMPANY TYPE] with [HEADCOUNT] employees in [LOCATIONS], first page only, and save the results to a table called "ICP scoring". 2) Score everyone 1 to 10 from headline and company against this ICP: [YOUR ICP], with a one-line reason. Tag 7 and above as "fit" and the rest as "rejected". 3) Open the full profiles of the top 3 only and confirm the fit. 4) Show a short table: first name, last initial, title, company, score, reason. 5) Write a connection note under 200 characters for each of the top 3, built from their role and company. Sound like a person, no pitch, no product name, no dashes.
Then
Send these connection requests now and show their status.
Example output fields
linkedin_url, full_name, headline, company, location, icp_score, reason, connection_note, status
// Connects with
Integrations used
// Fits into
Use cases
// Questions
FAQ
- Why score before outreach?
- Daily sending capacity is limited. Scoring first puts every request and every personal note on people who can actually buy.
- How does the AI score?
- It reads the profile and returns a 1 to 10 score with a short reason, based on the ICP you describe in plain words.
- What happens to rejected leads?
- They stay in the table with their score and reason. Read them once a week: they show exactly where your ICP description is too loose or too strict.
- How is this different from Clay?
- Clay charges to enrich and score every row. Here free rules filter first and AI only scores the survivors, which makes it up to 20X more cost-effective for LinkedIn enrichment.
// Explore more
Related playbooks
Enrich any LinkedIn profile with full structured data
Anyone writing personalized outreach or scoring leads on more than a headline. Search results give you a name and a headline. Outreach built on a headline sounds generic, and copying profile fields by hand stops scaling after a dozen leads.
Turn hiring announcements into outbound leads
Teams that sell to companies building or scaling a sales function: sales tools, agencies, RevOps services. A company hiring SDRs or a RevOps manager has a new priority, a budget and a pain right now. But the signals are scattered across posts and job pages, many come from recruiting agencies, and the person who posts the role is usually not the buyer. By the time someone checks by hand, the signal is stale.
Turn LinkedIn post likers and commenters into qualified leads
Founders, GTM engineers and outbound agencies who want warm leads instead of cold lists. People who like and comment on posts in your niche are raising their hand: the topic is on their mind this week and they will recognize what you talk about. But the signal is scattered across hundreds of posts, and working it by hand means opening every post, scrolling every reaction and reading every profile. So most teams go back to bought lists that accept 10 to 20% of requests.
// Get started
Start building with AtomSell Actions
Create an account, get an API key, and run this playbook end-to-end.
Create your account