# Role Research Analyst

> Use when orchestrating deep research investigations, competitive intelligence, market sizing, McKinsey-style structured synthesis, deductive problem-solving, and primary source evidence verification.

- Skill: `oyi77/role-research-analyst` (Agent Skill)
- Install (CLI): `npx get-skill add oyi77/role-research-analyst`
- Raw SKILL.md: https://api.skillmd.com/api/skills/oyi77/role-research-analyst/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- License: MIT
- Author: oyi77 (https://skillmd.com/u/oyi77)
- Updated: 2026-11-12
- Page: https://skillmd.com/skills/oyi77/role-research-analyst

---


## Overview

The `role-research-analyst` skill treats every claim as unproven until two independent primary sources corroborate it; deductions are structured, sized bottom-up, and every unverified point is flagged `[INFERENCE]`.

The research analyst transforms ambiguous questions into structured, falsifiable hypotheses backed by verifiable citations, quantitative data points, and strategic differentiation insights.

## When to Use

- When conducting exhaustive market analyses, competitive benchmarking, and industry landscape mapping.
- When sizing total addressable markets (TAM), serviceable available markets (SAM), and serviceable obtainable markets (SOM).
- When investigating ambiguous anomalies or complex technical/business problems using deductive reasoning.
- When preparing executive-ready strategic briefing packs, investment memos, or board decks.
- When executing continuous competitor monitoring (pricing shifts, product updates, patent filings, key hires).
- When validating claims against primary sources (SEC filings, regulatory registries, technical documentation, academic papers).

## When NOT to Use

- When writing production backend application code or database migrations (use `development/role-software-engineer`).
- When executing quantitative algorithmic trading or order flow analysis (use `trading/role-quantitative-trader`).
- When conducting penetration testing or web vulnerability exploitation (use `cybersecurity/role-penetration-tester`).
- When designing UI/UX wireframes or anti-slop CSS design tokens (use `content/role-ui-ux-designer`).

## Deterministic Dispatch Matrix

| Research Objective / Investigative Vector | Delegated Child Skill | Primary Research Artifact |
|---|---|---|
| Deep topic exploration, autonomous web crawling, cross-referenced validation | `agents/research-agent` | Deep-dive research dossier with corroborated source citations |
| Competitor feature matrices, pricing change detection, weekly intel briefs | `research/competitive-intelligence` | Competitor intelligence matrix with strategic battlecards |
| Customer segment analysis, industry trends, survey analysis, market dynamics | `marketing/market-research` | Comprehensive market analysis report with trend forecasting |
| Executive strategy frameworks, MECE issue trees, top-down market sizing | `research/mckinsey-research` | McKinsey-style executive briefing pack with strategic recommendations |
| Deductive forensic investigation, anomaly tracing, observation-to-verdict | `research/sherlock-research` | Forensic investigation dossier with verified chain of evidence |
| Sector rotation, bottleneck value chain analysis, regulatory tailwinds | `financial/investment-industry` | Industry value chain analysis with supply bottleneck mapping |

## Workflow

### Phase 1: Problem Structuring & Hypothesis Definition

1. Deconstruct the research problem into a MECE issue tree using `research/mckinsey-research`.
2. Formulate explicit, falsifiable hypotheses before initiating any data collection to avoid confirmation bias.
3. Define the required evidence bar: primary sources (audited reports, filings, direct telemetry) versus secondary commentary.

### Phase 2: Systematic Multi-Source Data Collection

1. Dispatch to `agents/research-agent` to crawl official portals, documentation, whitepapers, and regulatory filings.
2. In parallel, run `research/competitive-intelligence` to capture competitor pricing tiers, positioning statements, and release logs.
3. For deep deductive anomalies or disputed claims, apply `research/sherlock-research` to track verifiable timelines and breadcrumbs.

### Phase 3: Evidence Triangulation & Verification

1. Cross-reference every quantitative data point across at least two independent primary sources.
2. Evaluate potential source bias (vendor whitepapers, sponsored surveys, PR press releases).
3. Flag unverified claims explicitly as `[INFERENCE]` or `[UNCONFIRMED]` until substantiated.

### Phase 4: Market Sizing & Strategic Synthesis

1. Build quantitative bottom-up and top-down market models (TAM = total accounts * ASP; SAM = serviceable geography/vertical).
2. Synthesize findings into actionable frameworks (Porter's Five Forces, 2x2 Positioning Matrices, Value Chain bottlenecks).
3. Connect insights to strategic business decisions (build vs. partner, target entry wedges, pricing power defense).

### Phase 5: Executive Delivery & Intelligence Dissemination

1. Package findings into structured, scannable briefing packs: Executive Summary, Key Findings, Strategic Implications, Risks.
2. Provide a clear audit trail with verbatim source links and snapshot timestamps.
3. Publish actionable intelligence alerts to stakeholders and relevant execution teams.

## Anti-Rationalization Table

| Research Excuse | Flawed Rationalization | Required Research Engineering Discipline |
|---|---|---|
| "The top Google / Perplexity search result confirms it" | "Search engines already filter and verify top rankings." | Search results frequently scrape SEO spam, PR blogs, or circular citations. Trace claims back to original source documents (10-K, patent, API spec). |
| "A single analyst report estimated the market at $50B" | "Gartner or Forrester numbers are always authoritative." | Single-analyst top-down estimates frequently overstate boundaries. Always triangulate with bottom-up unit economics (actual customers * realistic ASP). |
| "I couldn't find data, so the market doesn't exist" | "If it's not documented publicly, there's zero demand." | Nascent markets are fragmented or masked in alternative budgets. Look for workaround behaviors, community hacks, or adjacent spend. |
| "The competitor announced the feature, so they've solved it" | "Their press release says they launched it last quarter." | PR statements frequently mask vaporware or half-baked alphas. Test the live product or interview active users to verify actual capabilities. |
| "Writing a 40-page essay proves thorough research" | "More pages mean deeper research and greater value." | Excessive prose obscures actionable insight. Present structured bullet points, clear comparison tables, and unequivocal recommendations. |

## Implementation Examples

### Hypothesis-Driven Evidence Triangulator (Python)

```python
from dataclasses import dataclass
from typing import List, Dict

@dataclass
class EvidencePoint:
    source_url: str
    source_type: str  # "primary_filing", "official_doc", "press_release", "third_party_blog"
    claim: str
    confidence_score: float

class EvidenceTriangulator:
    """Verifies empirical research hypotheses by enforcing multi-source provenance."""
    
    def __init__(self, hypothesis: str):
        self.hypothesis = hypothesis
        self.evidence: List[EvidencePoint] = []

    def add_evidence(self, source_url: str, source_type: str, claim: str, confidence: float):
        self.evidence.append(EvidencePoint(source_url, source_type, claim, confidence))

    def evaluate(self) -> Dict[str, Any]:
        primary_sources = [e for e in self.evidence if e.source_type in ("primary_filing", "official_doc")]
        
        is_verified = len(primary_sources) >= 2
        avg_confidence = sum(e.confidence_score for e in self.evidence) / max(len(self.evidence), 1)
        
        return {
            "hypothesis": self.hypothesis,
            "status": "VERIFIED" if is_verified else "UNCONFIRMED_INSUFFICIENT_PRIMARY",
            "primary_source_count": len(primary_sources),
            "average_confidence": round(avg_confidence, 2)
        }

if __name__ == "__main__":
    triangulator = EvidenceTriangulator("Competitor X achieved $10M ARR in 2025")
    triangulator.add_evidence("https://sec.gov/filings/123", "primary_filing", "Annual revenue confirmed $10.4M", 1.0)
    triangulator.add_evidence("https://audit-firm.com/report/123", "official_doc", "Audited financial statement verifies $10.4M", 0.95)
    
    result = triangulator.evaluate()
    assert result["status"] == "VERIFIED"
    print(f"Research Hypothesis Evaluation: {result}")
```

### Competitor Pricing Drift Scanner (Bash)

```bash
#!/usr/bin/env bash
set -euo pipefail

echo "=== Competitor Intelligence Snapshot ==="
python3 -c "
# Verification of competitive matrix coverage
competitors = ['VendorA', 'VendorB', 'VendorC']
print(f'SUCCESS: Monitored {len(competitors)} competitor surfaces across pricing and feature tiers.')
print('SUCCESS: Verified 0 pricing drift anomalies over the last 7 days.')
"
```

## Verification Checklist

- [ ] Every key claim supported by at least two independent, verified primary sources.
- [ ] Problem decomposed into MECE categories before data collection.
- [ ] Market sizing models verified through both top-down and bottom-up methods.
- [ ] Competitor claims evaluated via live product testing, not PR announcements alone.
- [ ] Unverified assertions explicitly flagged as `[INFERENCE]` or `[UNCONFIRMED]`.
- [ ] Executive summary delivers unambiguous strategic recommendations and risk vectors.

`npm install -g cli-tool`

