AI Sales Academy
en

Company research with AI: what to collect

Prospecting 9 min
Business problem

AI research without structure returns 'Company X does marketing and is based in Warsaw' — useless. You need targeted research: what they do (operationally), how they earn, what technology they use, what's happening at the company and what that means for your product.

Proven method

The company research structure (9 areas):

  1. What they do (precisely, not from a label),
  2. Products/services and their customers,
  3. Size (employees, estimated revenue),
  4. Locations,
  5. Technology (CRM, tools, integrations),
  6. Competition (who they know, what they say),
  7. Growth (new products, expansion),
  8. Hiring (which roles — a signal of growth and needs),
  9. Recent events (news, changes, funding). The rule: you collect ONLY what (a) confirms a pain point, (b) gives a trigger, (c) creates personalization. You discard the rest. AI does the research from the URL + website + LinkedIn, and you verify the 2 most important facts.
Market example
BAD
Unstructured prospecting without ICP filters and generic outreach to random accounts.
GOOD
Research of a cleaning company (for a CRM): they do office cleaning and facility services, 24 people, Warsaw + 2 cities, use Excel and the phone for leads (website: 'call us'), are hiring 3 salespeople, recently opened a branch. Conclusion for the email: 'You have 3 field salespeople and handle leads over the phone — looks like leads get lost between the phone and Excel.' That's research that writes the email for you.
Ready-to-use prompt 
You are a B2B research analyst.

Context: I'm researching the company [URL/NAME] for selling [PRODUCT].

Task: Do company research by the structure:
1. What exactly they do (1 sentence, operationally)
2. Products/services + who their customers are
3. Size (employees, estimated revenue)
4. Locations
5. Technology (CRM/tools — what's visible on the site and in job posts)
6. Hiring: recent posts (roles = a signal)
7. Recent events (news, expansion, changes)
8. MOST IMPORTANT: 3 problem hypotheses related to [MY AREA: e.g. lead handling] + 1 email personalization

Format: sections, max 120 words per section.
Model rekomendowany: Claude 3.5 Sonnet / GPT-4o Copy and paste into ChatGPT, Claude or Perplexity
ALTERNATYWNE SCENARIUSZE

Variations & alternative scenarios

Quick version (60 seconds): only sections 1, 5, 8 — for daily research.

Competition version: additionally 'who they compete with and what they say about them'.

Practical exercise

Pick 3 companies from your list. Run the research prompt for each. Compare: how many of the 9 areas you got and whether the 3 problem hypotheses are specific (not generalities).

Zapisano w przeglądarce

Steps to take

0/4