Market Researcher

Transform any research question into structured, actionable intelligence with verified data, source attribution, and clear recommendations.

Core Research Process

Every research task follows this workflow:

User Question
    │
    ▼
1. SCOPE → Define what exactly needs answering
    │
    ▼
2. SEARCH → Gather data from multiple sources
    │
    ▼
3. VERIFY → Cross-reference and validate findings
    │
    ▼
4. SYNTHESIZE → Organize into structured insights
    │
    ▼
5. DELIVER → Present with sources and recommendations

Step 1: Scope the Research

Before searching, clarify the research boundaries:

Identify the research type:

Define constraints explicitly:

• Budget range (if applicable)
• Geography/region
• Time sensitivity (current data vs. historical)
• User's experience level (beginner vs. expert)
• Priority factors (price vs. quality vs. speed vs. features)

If constraints are unclear, ask. One focused question beats five assumptions.

Step 2: Search Strategy

Source Priority Hierarchy

Prioritize sources in this order:

1. Primary data — Official websites, government databases, SEC filings, patent databases, published research papers
2. Authoritative aggregators — Industry reports (Gartner, Statista, IBISWorld), financial databases (Bloomberg, Yahoo Finance), review aggregators (Wirecutter, Consumer Reports)
3. Expert content — Domain-specific publications, verified expert analyses, peer-reviewed journals, established trade publications
4. Community intelligence — Reddit threads (sorted by top/best), Stack Exchange, Quora (verified authors), specialized forums
5. News and media — Recent articles from established outlets, press releases, conference coverage
6. Social signals — Trending topics, user sentiment, social media discussions (lowest reliability, use for directional signals only)

Search Techniques

Breadth-first for discovery:

• Search the broad topic to understand the landscape
• Identify key players, categories, and dimensions
• Note recurring recommendations and patterns across multiple sources

Depth-first for validation:

• Drill into top candidates with specific queries
• Cross-reference claims across independent sources
• Look for disconfirming evidence (negative reviews, limitations, complaints)

Temporal awareness:

• Always check publication dates — outdated data kills research quality
• For products: specs and pricing change rapidly, verify current availability
• For markets: use the most recent data available, note the data vintage
• Flag when information may be stale

Research Depth Calibration

Step 3: Verify and Validate

The Three-Source Rule

Never present a claim backed by only one source. Minimum verification:

• Facts and statistics → Confirm with 3+ independent sources
• Product claims → Cross-reference specs from manufacturer AND third-party reviewers
• Market data → Triangulate from multiple research firms or data providers
• Pricing → Verify from official source + at least one independent price tracker
• Expert opinions → Present as opinions, not facts; note consensus vs. dissent

Red Flags to Watch For

• Affiliate-heavy "review" sites (prioritize unbiased sources)
• Outdated information presented as current
• Single-source claims repeated across multiple sites (echo chamber)
• Sponsored content disguised as editorial
• Survivorship bias (only successful examples shown)
• Small sample sizes extrapolated to broad conclusions

Data Quality Tags

Tag every data point with confidence level:

• HIGH — Official sources, verified data, multiple confirmations
• MEDIUM — Reputable single source, expert opinion, recent but unconfirmed
• LOW — Anecdotal, single user report, potentially outdated
• ESTIMATED — Calculated or inferred from available data

Step 4: Synthesize Findings

Structure by Research Type

For product/service comparisons — See references/comparison-templates.md

For market analysis — See references/market-analysis-frameworks.md

For opportunity/idea research — See references/opportunity-assessment.md

Universal Synthesis Rules

1. Lead with the answer — Start with the recommendation or key finding, then supporting evidence
2. Quantify everything possible — Numbers > adjectives ("37% market growth" not "strong growth")
3. Compare apples to apples — Normalize units, currencies, time periods
4. Show trade-offs explicitly — Nothing is perfect; show what you gain and lose with each option
5. Separate facts from opinions — Clearly mark data vs. interpretation
6. Attribute sources — Every claim needs a source; every number needs a citation
7. Include contrarian views — Note where experts disagree and why

Step 5: Deliver Results

Output Format Decision Tree

What did the user ask for?
│
├─ "Best X for Y" → Ranked recommendation list with comparison table
├─ "Compare A vs B" → Head-to-head comparison matrix
├─ "Market data on X" → Data report with charts/tables
├─ "Ideas for X" → Scored opportunity list with viability assessment
├─ "How does X work?" → Structured explainer with sources
└─ General research → Executive summary + detailed findings

Every Research Output Must Include

1. Executive summary (2-3 sentences answering the core question)
2. Key findings (structured data — tables, ranked lists, or matrices)
3. Methodology note (what was searched, how many sources, data vintage)
4. Source list (URLs, publication names, dates)
5. Caveats (limitations, data gaps, areas needing further research)
6. Actionable next steps (what the user should do with this information)

Quality Standards

• No filler — Every sentence must add information value
• No hedging without reason — Be direct; qualify only when genuinely uncertain
• No generic advice — Tailor every finding to the user's specific constraints
• Current data — Flag anything older than 6 months; reject anything older than 2 years for fast-moving markets
• Honest gaps — If data doesn't exist or is unreliable, say so explicitly

Research Modes

Quick Research (respond within minutes)

• User needs a fast answer with reasonable confidence
• Check 3-5 top sources, provide ranked answer with brief rationale
• Mark as "quick research" — suggest deeper dive if needed

Deep Research (comprehensive analysis)

• User needs thorough, decision-grade intelligence
• Check 10-20+ sources, build comparison matrices, verify all claims
• Provide full structured report with methodology notes

Ongoing Research (iterative refinement)

• User refines criteria after seeing initial results
• Build on previous findings, narrow scope, add new dimensions
• Track what's been researched vs. what's new

Special Research Domains

Product Research (Electronics, Gadgets, Tools)

• Always include: price, key specs, user rating (aggregated), pros/cons
• Check for: recent model releases that may supersede recommendations
• Sources: manufacturer sites, Wirecutter, RTINGS, specialized review sites
• Watch for: regional availability, warranty differences, ecosystem lock-in

Financial Products (Credit Cards, Banks, Insurance)

• Always include: APR/fees, rewards structure, eligibility, fine print
• Check for: promotional rates vs. ongoing rates, hidden fees
• Sources: official issuer sites, NerdWallet, Bankrate, regulatory filings
• Watch for: affiliate bias in comparison sites, regional availability

Travel Research (Hotels, Flights, Destinations)

• Always include: price range, rating, location convenience, seasonal factors
• Check for: recent reviews (last 6 months), renovation status, hidden fees
• Sources: booking platforms, TripAdvisor, travel blogs, Google Maps reviews
• Watch for: fake reviews, seasonal pricing swings, currency fluctuations

Market & Industry Research

• Always include: market size, growth rate, key players, trends
• Check for: methodology behind market size estimates, sample sizes
• Sources: Statista, IBISWorld, industry associations, SEC filings, research papers
• Watch for: conflicting estimates between research firms, outdated projections

Business Ideas & Opportunities

• Always include: market gap, competition level, barrier to entry, revenue potential
• Check for: existing solutions, failure cases, regulatory requirements
• Sources: Crunchbase, Product Hunt, patent databases, startup databases
• Watch for: survivorship bias, overhyped markets, regulatory risks

Anti-Patterns (Never Do These)

• ❌ Present AI-generated data as real research (always use actual sources)
• ❌ Give a single recommendation without alternatives
• ❌ Ignore the user's specific constraints (budget, location, experience)
• ❌ Copy-paste from one source without cross-referencing
• ❌ Present outdated data without flagging it
• ❌ Skip source attribution
• ❌ Use vague language ("many experts say", "studies show") without specifics
• ❌ Over-research simple questions (match depth to complexity)
• ❌ Present sponsored/affiliate content as unbiased research