n8n Workflow: Research Leads & Draft Outreach w/ Anthropic

Research leads and draft personalized cold outreach—automatically

This n8n workflow researches a company from its website using an Anthropic chat model and an HTTP page-fetching step, then returns a structured company profile, an ICP fit verdict (GOOD/MAYBE/POOR), and a draft outreach message—either in n8n Chat or written back to Google Sheets in bulk.

What this workflow does

• Starts when you send a message in n8n Chat or run it manually to process a Google Sheets list of domains.
• Loads your configuration including your seller profile, ideal customer definition, and limits such as the maximum number of pages to read.
• Bulk mode formatting: reads rows from Google Sheets, normalizes each domain to a standard https:// URL, and creates a research prompt per company.
• AI-led page selection & extraction: uses an Anthropic chat model with short-term conversation memory to decide which pages to fetch and what information to extract.
• Iterates through public pages over HTTP, converts HTML to readable text, and reviews key pages (e.g., About, Pricing, Careers, Blog) up to your page limit.
• Outputs lead intelligence: signals, fit verdict + fit score and rationale, pain hypotheses, questions to ask, unknowns, and a personalized opening message.
• Writes results to Google Sheets when the input came from the spreadsheet, updating the matching row with generated fields.

Use cases

• Qualify SaaS leads by determining ICP fit from public website signals
• Generate personalized cold outreach drafts at scale from a domains list
• Speed up sales research by summarizing key pages and surfacing unknowns

Technical details

• Nodes/stack mentioned: if, set, code, sticky note, Google Sheets, manual trigger
• Core capabilities: Anthropic chat model, short-term conversation memory, HTTP page fetching, HTML-to-text extraction