AI Sales Academy
en

Company sources and AI-generated search queries

Prospecting 12 min
Business problem

Manual company search is dozens of queries that still return the same 20 results. AI knows hundreds of phrasing variants, domains, directory sites and can build boolean searches — but only if you give it a concrete ICP and the task 'generate queries', not 'find companies'.

Proven method

Company sources (10): Google, Google Maps, LinkedIn, industry directories, marketplaces, business databases (CEIDG/KRS), social media, job portals (posts = active companies), review portals (G2, Capterra, Google reviews), your own data. AI helps in 5 ways:

  1. generates search queries in language variants,
  2. builds keyword combinations (product + industry + location + problem),
  3. points out directories for the niche,
  4. creates boolean searches (site: linkedin.com/company + phrases),
  5. expands the list: from 20 found companies it generates 100 similar ones by pattern. The rule: AI generates QUERIES and PATTERNS, and you (or tools) run the searches.
Market example
BAD
Unstructured prospecting without ICP filters and generic outreach to random accounts.
GOOD
Instead of 'marketing agencies Warsaw' AI creates a set: marketing agency Warsaw, performance marketing Warsaw, SEO agency Warsaw, site:linkedin.com/company marketing agency Warsaw, digital agency Warsaw + a directory site (e.g. a company catalog). Each query is a separate list of potential companies, and from one good company AI generates 20 similar patterns.
Ready-to-use prompt 
You are a B2B search expert.

Context: I'm looking for companies to prospect. ICP: [SHORT DESCRIPTION: industry, size, location, stack]. Niche: [NICHE].

Task: Generate a set of queries:
1. 10 search queries (different variants: industry name, problem, service, location)
2. 5 in English (if companies use English names)
3. 5 boolean searches (site:, AND, OR, NOT)
4. 3 directories typical for this niche
5. 5 example companies (if you know the niche) + 10 variants of 'similar' companies

Format: sections, ready to copy.
Model rekomendowany: Claude 3.5 Sonnet / GPT-4o Copy and paste into ChatGPT, Claude or Perplexity
ALTERNATYWNE SCENARIUSZE

Variations & alternative scenarios

Local services version: add Google Maps as a source + phrases like 'near me', cities.

SaaS version: add stack queries (site: LinkedIn 'uses HubSpot') and review portals.

Practical exercise

Take your niche and generate a set: 10 PL queries, 5 boolean searches, 3 directories. Run 5 queries and write down which gave the best companies.

Zapisano w przeglądarce

Steps to take

0/4