Research Methodology
Structured approach to finding, vetting, and synthesizing information from diverse sources. Turns research questions into trustworthy, actionable findings through systematic query design, source evaluation, and cross-referencing.
When to Use This Skill
- Conducting competitive analysis or market scans
- Investigating historical events, trends, or technical evolution
- Fact-checking claims across multiple sources
- Synthesizing research into structured deliverables (reports, tables, timelines)
- Any research task that requires more than a single search query
Quick Reference
| Resource | Purpose | Load when |
|----------|---------|-----------|
| references/search-strategies.md | Query design, source vetting, fact verification, synthesis techniques | Starting any research task |
Workflow
Phase 1: Scope → Define research objective, key questions, constraints
Phase 2: Explore → Design queries, search broadly, capture sources
Phase 3: Verify → Vet sources, cross-reference claims, assess credibility
Phase 4: Synthesize → Organize findings into structured deliverables
Phase 1: Scope the Research
Before searching, clarify the research objective:
- State the question -- what exactly are we trying to learn?
- Define success criteria -- what does a complete answer look like?
- Set constraints -- time period, geography, domains, source types
- List hypotheses -- what do we expect to find? (helps detect bias)
- Identify key terms -- domain vocabulary, synonyms, related concepts
Scoping Template
**Research Question**: [precise question]
**Success Criteria**: [what constitutes a complete answer]
**Constraints**: [time period, scope, source types]
**Key Terms**: [domain vocabulary and synonyms]
**Initial Hypotheses**: [what we expect, to check against later]
Phase 2: Explore
Design multiple query variations and search broadly before narrowing:
- Create 3-5 query variations per research question
- Search broadly first -- cast a wide net with general terms
- Refine iteratively -- narrow based on initial results
- Track what you searched -- record every query for reproducibility
Query Design Principles
- Use exact-match phrases in quotes for precision
- Exclude noise with negative keywords
- Target specific timeframes for recency or historical depth
- Vary terminology across queries to avoid vocabulary bias
- Use domain-specific operators when available (site:, filetype:, etc.)
Source Capture
For each promising source, record:
- URL and access date
- Key claims with direct quotes
- Author/publisher and their domain authority
- Any noted biases or limitations
Phase 3: Verify
Vet sources and cross-reference claims before trusting them:
- Assess source authority -- who wrote it, what are their credentials?
- Check recency -- is the information current enough for the question?
- Detect bias -- does the source have a commercial, political, or ideological interest?
- Triangulate -- require 2+ independent sources for any key claim
- Seek primary sources -- follow citation chains to the original data
Confidence Rating
| Level | Criteria | |-------|----------| | Confirmed | 3+ independent, authoritative sources agree | | Likely | 2 sources agree, no contradictions found | | Uncertain | Single source or sources disagree | | Contested | Credible sources directly contradict each other |
Phase 4: Synthesize
Organize findings into a structured deliverable:
Standard Research Report Structure
## Research Summary
[1-2 paragraph overview of findings]
## Key Findings
- [Finding 1] — [confidence level]
- [Finding 2] — [confidence level]
## Detailed Analysis
[Organized by theme or question]
## Source Credibility Assessment
| Source | Authority | Recency | Bias Risk | Rating |
|--------|-----------|---------|-----------|--------|
## Gaps and Limitations
[What we couldn't determine and why]
## Recommendations
[Next steps or actions based on findings]
Anti-Patterns
- Do not rely on a single source for any key claim
- Do not present uncertain findings as confirmed facts
- Do not skip source vetting for convenience
- Do not omit contradictory evidence -- always surface disagreements
- Do not let initial hypotheses bias which findings you report