Enrichment: data and the verification rule
Prospecting 9 min
0/28
Business problem
AI is great at guessing emails by pattern (first.last@company.com), but it also guesses when it doesn't know — it returns contacts that don't exist. Sending 500 emails to unverified addresses = 10-30% bounce rate, a burned domain, zero replies. Contact data requires verification at the source.
Proven method
Enrichment in 3 layers:
- Basic data (company, website, industry, location, employees) — AI + sources (LinkedIn, website, registries),
- Contact data (name, role, email, LinkedIn, phone) — TOOLS + verification (Hunter, Apollo, Snov, Clearbit; SMTP verification), AI is never the only source,
- Context data (technology, funding, hiring, revenue, growth, intent) — AI analyzes public data. The rule 'AI = hypothesis, Source = verification': every fact (especially email and phone) must come from a tool or be verified. You use AI for completion and connection, not for creating truth.
Market example
- BAD
- Unstructured prospecting without ICP filters and generic outreach to random accounts.
- GOOD
- AI suggests 'jan.kowalski@company.com' (pattern). Verification in Hunter: exists with 90% confidence. But AI also suggests 'anna.nowak@company.com' — verification: nothing, 0 results. You don't send. Contextually AI adds: the company hires 3 salespeople (job posts), uses WordPress (site analysis), traffic grew 40% (SimilarWeb). These context facts are fine from AI — because they don't risk bounces.
Ready-to-use prompt
You are a B2B data consultant. Context: I'm enriching a lead list. Product: [PRODUCT]. ICP: [SUMMARY]. Task: Design the enrichment process: 1. List the basic, contact and context data worth collecting (3 layers) 2. For each: where to get it (AI / tool / public source / SMTP verification) 3. Point out which data AI can safely guess and which MUST be verified 4. Propose a spreadsheet/CRM structure: columns for a lead with this data Format: 4 sections, tables where possible.
Model rekomendowany: Claude 3.5 Sonnet / GPT-4o Copy and paste into ChatGPT, Claude or Perplexity
ALTERNATYWNE SCENARIUSZE
Variations & alternative scenarios
Small-budget version: free tools (Hunter free, Google, LinkedIn) + AI research.
Large-scale version: enrichment API tools + dedup + bulk validation.
Practical exercise
Take 10 leads from your list. For each: write down which contact data you have verified and which is 'guessed'. Create a spreadsheet with columns: company, email (status: verified/guess), LinkedIn, phone, technology, hiring, intent.
Zapisano w przeglądarce