banatie-strategy/research/passive-research-methodolog...

5.5 KiB

Banatie Passive Research: Methodology & Workflow

Created: November 1, 2025
Purpose: Guide for validating ICP hypotheses through passive research
For: Future research sessions with @men in new chats
Status: Battle-tested methodology


🎯 WHY Passive Research Before Interviews

Problem with going straight to interviews:

  • Ask wrong questions → waste people's time
  • Get polite, filtered answers → miss real pain
  • Can't challenge assumptions → build wrong thing

Passive research gives us:

  • Real pain points (people complain authentically online)
  • Actual language people use (for messaging)
  • Competitive intel (what exists, what fails)
  • Confidence before talking to customers

"Don't ask people what they want. Observe what they complain about when they think you're not listening."

Flow: Passive research → Interview validation → Build confidence → Launch


🛠 Tools & Resources

Search Tools (MCP):

  1. Brave Search MCP - Main search engine for Reddit/forums
  2. Perplexity MCP - Secondary for synthesis

Project Knowledge Files:

  1. banatie-research-directions-tracker.md - Core research directions (1-10)
  2. banatie-research-directions-tracker-extension1.md - Additional research directions
  3. Completed research - direction-1-strong-signals.md, direction-6-strong-signals.md, etc. (for comparison)

Important: Previous research files will be in Project Knowledge for reference.


🔄 The Workflow (Step-by-Step)

Phase 1: Session Start

@men loads context:

  • Read: ROADMAP.md, INDEX.md, 12-the-current-tech-state.md (in /projects/my-projects/banatie-strategy/)
  • Read: banatie-research-directions-tracker.md and banatie-research-directions-tracker-extension1.md (Project Knowledge)
  • Review: Completed research files
  • Confirm with Oleg: "Ready to start Direction N - [hypothesis]"

Phase 2: Planning Searches (Collaborative)

Pattern:

  • Oleg: "Let's search Direction N" OR "Also try X"
  • @men: Proposes 5-7 specific search queries
  • Discuss → Adjust → Execute

Key: DISCUSS before searching, ADJUST on the fly

Phase 3: Execute Searches (Iterative)

Pattern:

  1. Run search → Quickly scan → Share findings immediately
  2. Discuss: "Dig deeper?" OR "Move on?" OR "Try different angle?"
  3. Adjust strategy based on findings

Important: Multiple search variations per topic (3-5 variations to avoid false negatives)

Phase 4: Document Findings (Continuous)

Create downloadable file in chat: direction-N-strong-signals.md

Format:

# Direction N Research: Strong Signals Found

## 🟢 STRONG SIGNAL #1: [Title]
- **URL:** [exact link]
- **Quote:** "[exact quote]"
- **Context:** [why this matters]

## 🟡 MEDIUM SIGNAL #2: ...
## 🔴 WEAK/MISSING SIGNAL #3: ...

Update as you go - after every 2-3 searches using str_replace

Phase 5: Discussion & Decisions

After 5-7 searches OR when pattern emerges:

  • @men summarizes: Found X signals, key insight, unexpected findings
  • Oleg decides: Keep searching / Move to next direction / Pivot

Key: COLLABORATIVE decision-making

Phase 6: Wrap-Up

@men creates:

  • Full research doc (English): direction-N-strong-signals.md

Oleg uploads to Project Knowledge for future sessions


🎨 The "Vibe" - How We Actually Work

Speed & Iteration:

  • Discuss → Search → Find → Discuss → Adjust → Search again
  • Real-time collaboration, quick pivots

Collaborative Discovery:

  • Oleg challenges, adds ideas, spots patterns
  • @men proposes strategy, interprets, suggests pivots
  • Together: hunting for truth, not confirming biases

Hypothesis Testing:

  • We WANT to be wrong (better to kill bad ideas in research)
  • Try to DISPROVE hypotheses
  • Celebrate negative findings

Documentation:

  • Document EVERYTHING (signals, absences, questions)
  • Exact quotes with URLs (proof)
  • Note what's MISSING (absence = evidence)

📋 Quality Standards

Good Search Queries:

  • ✅ Specific: site:reddit.com/r/webdev "client approval" placeholder images
  • ❌ Too broad: website images problems

Strong Signal Indicators:

  • 🟢 Multiple threads (3+ sources)
  • 🟢 Specific examples, emotional language
  • 🟢 Solutions attempted but failed
  • 🟢 Willingness to pay mentioned

Weak Signals:

  • 🔴 Single mention, vague complaints
  • 🔴 No urgency, theoretical discussion

🚀 Starting a New Session

@men in new chat:

  1. Read this methodology
  2. Load: ROADMAP.md, INDEX.md, tracker, completed research
  3. Confirm: "Ready to start Direction N - [hypothesis]"

Oleg:

  1. Share this document
  2. Wait for @men to load context
  3. Start researching

📊 Success Metrics

Good session:

  • ✅ Found 2-3 strong signals with proof
  • ✅ Validated OR invalidated hypothesis
  • ✅ Generated interview questions
  • ✅ Clear decision: continue / pivot / kill

Poor session:

  • ❌ Generic findings, no proof
  • ❌ Confirmation bias
  • ❌ Vague conclusions

🌟 Core Philosophy

"We're not trying to prove we're right. We're trying to discover what's true."

Goal: Find REAL problems → Validate quickly → Build confidence → Make data-driven decisions

Success = Truth discovered, whether it supports our idea or not.


Version: 1.0
Last Updated: November 1, 2025