lead-qualification

Research and qualify business leads against your ICP

SKILL.md

GitHub

Lead Qualification

Research a lead or list of leads to determine ICP fit, company details, and deal potential.

Usage

/lead-qualification "John Smith john@company.com" — qualify a single lead /lead-qualification — then paste a list of leads

Works with

Runs on public web research alone. If a Cogny workspace is connected, it adds first-party signals; call mcp__cogny__list_integrations once to see what's there. If it isn't, skip the connected rows and say what connecting adds.

Job Primary Also answers it What it adds
Company, role, size WebSearch / WebFetch — Always available
Already in the CRM? HubSpot search_crm_objects Upsales list_contacts / list_accounts; Pipedrive list_deals(org_id) once you know the org Existing customer, open deal or past loss changes the next step
Engaging with our ads? Fibbler get_company_engagement — LinkedIn impressions, clicks and engagements by company

Pipedrive has no search tool here; only check it when the user gives an org id or the list is small enough to page list_organizations.

Steps

0. Pin down the ICP

Scoring needs a target. If the user hasn't given one, ask: "Who is your ideal customer (industry, company size, region, buyer role, must-have signals like 'runs paid media')?" If they'd rather not spell it out, infer a draft from their own site: WebFetch their homepage, pricing, customers/case-studies pages, and note who the copy, logos and pricing tiers address. Show the inferred ICP in 3–5 bullets and get a yes or corrections before scoring. Score every lead against that confirmed ICP and repeat it at the top of the output.

1. Parse the lead(s)

Extract name, email, and company from input. If email domain is personal (gmail, hotmail, outlook, yahoo, live), flag it — personal emails are a lower-quality B2B signal.

2. Research each lead

For each lead, WebSearch for:

  • Full name + company → LinkedIn profile, role, seniority
  • Company domain → website, size, industry
  • Company + "marketing" or "advertising" → whether they do paid media

3. Assess company fit

WebFetch the company website if found:

  • What do they sell?
  • How big are they? (team page, about page, funding)
  • Do they run ads? Check for tracking pixels with curl (WebFetch summarises the page and drops <script> tags):
    curl -sL -A "Mozilla/5.0" "https://<company-domain>/" | grep -oiE "googletagmanager\.com/gtm\.js|gtag/js\?id=(AW|G)-[A-Z0-9]+|fbevents\.js|snap\.licdn\.com|px\.ads\.linkedin\.com|analytics\.tiktok\.com|static\.ads-twitter\.com|bat\.bing\.com|redditstatic\.com/ads|hs-scripts\.com" | sort -u
    
    AW- IDs mean Google Ads conversion tracking; fbevents.js = Meta Pixel; snap.licdn.com = LinkedIn Insight Tag; GTM alone hides what's inside it, so say "GTM present, ad pixels unknown". Also look for ad landing pages (/lp/, utm_ links).
  • What's their likely marketing budget?

3b. Enhanced when connected: first-party signals

If Cogny isn't connected, skip this and note what connecting adds (whether the lead is already a customer, has an open or lost deal, or has been engaging with your LinkedIn ads).

  • CRM match. HubSpot: mcp__cogny__mcp__HubSpot__search_crm_objects on contacts (filter email EQ the lead's email) and on companies (filter domain EQ the email domain). Upsales: list_contacts(filters: [{"field": "email", "comparator": "eq", "value": "<email>"}]), then list_orders(filters: [{"field": "client.id", "comparator": "eq", "value": <account id>}]) for history. Pipedrive: page mcp__cogny__mcp__Pipedrive__list_organizations and match on name/domain, then list_deals with that org_id. Report: existing customer, open deal, closed-lost (with reason if present), or not found.
  • Ad engagement (Fibbler). mcp__cogny__mcp__app-fibbler-co__get_company_engagement(company_search: "<company>", time_period: "90d") shows whether the company has been seeing and engaging with LinkedIn ads; get_influenced_deals tells you if it already sits in influenced pipeline. Treat engagement as a warmth signal, not intent.

Never write to the CRM from this skill. If the user wants the lead created or updated, tell them which record to change.

4. Score each lead

ICP: [confirmed ICP, one line]
Lead: [Name] ([email])
Company: [Company] — [what they do]
Role: [Title/seniority if found]
ICP Fit: HIGH / MEDIUM / LOW / NONE

Signals:
+ [Positive signal, e.g., "Runs Google Ads"]
+ [Another positive]
+ [First-party signal, e.g., "Fibbler: 1,240 LinkedIn impressions, 9 engagements in 90d"]
- [Negative signal, e.g., "Personal email"]
- [Another negative]

Estimated deal value: [based on company size]
CRM status: [Customer / Open deal / Lost (reason) / Not in CRM / Not checked]
Next step: [specific recommendation]

5. Summary table for multiple leads

┌──────────────┬──────────────────┬──────────┬───────────────┐
│     Name     │     Company      │ ICP Fit  │   Next Step   │
├──────────────┼──────────────────┼──────────┼───────────────┤
│ [Lead 1]     │ [Company]        │ HIGH     │ Send proposal │
│ [Lead 2]     │ [Unknown]        │ LOW      │ Disqualify    │
└──────────────┴──────────────────┴──────────┴───────────────┘