Generate B2B Persona Insights

This bundle assembles a complete research pipeline for investigating any B2B Ideal Customer Profile end to end. Once set up, giving it a customer profile description - a persona, company size, and industry - produces a structured report covering that persona's actual workflows, real pain points sourced from forums and discussions, cost structure (software and personnel expenses), and authentic quotes with source links, all in about 5 minutes from a single command. The underlying methodology runs a five-phase analysis: mapping the ICP's roles/KPIs/team structure, researching daily workflows and time allocation, surfacing real pain points with evidence, breaking down direct and indirect costs, and synthesizing key insights on market size and competitive landscape - powered by web search across LinkedIn/Reddit/forums/industry sites plus deeper scraping for niche or protected content that a basic search wouldn't reach.

It's been validated on real B2B verticals: a 22,000-word report on accountants at Cyprus SMBs ($500K-$3M revenue) covering salary data, tool costs, and workflow breakdowns; market sizing and competitor analysis for IT founders relocating to Cyprus; and screening-workflow and tool-limitation research for HR managers at agencies with 500+ employees.

Use this bundle when validating a market hypothesis before building a product, mapping a customer segment's actual workflows and pain points for product decisions, understanding a prospect's challenges and budget for sales conversations, or crafting marketing messaging grounded in real customer language rather than assumptions. It's built for Claude Code or Claude Desktop with MCP support specifically.

It requires at minimum a Tavily API key (a free tier works for basic web search); a Bright Data API key is optional but recommended for reaching niche forums and protected content a basic search can't access, and a Reddit API key is optional for direct Reddit search rather than scraped Reddit content. Without at least the Tavily connection, the research pipeline has no way to search the web at all.

Install the two MCP servers - Tavily Search and Bright Data - and configure their API keys. Copy the ICP Research Skill prompt into your Claude Code project instructions. Then run /icp-research <your ICP description> (e.g. "Accountants in Cyprus, companies $500K-$3M revenue") and receive a structured Markdown report within about 5 minutes, covering the ICP profile, workflow analysis, pain points, cost analysis, and key insights.

Founders validating market hypotheses before building, product managers mapping customer workflows and pain points, sales teams understanding prospect challenges and budgets, and marketing teams crafting messaging based on real customer language - anyone who needs an evidence-backed B2B persona profile rather than an assumption-based one.