Research

---
name: research
description: When the user wants competitive intelligence, audience research, market analysis, or needs to understand a market before building a campaign. Use when the user mentions "research," "competitor analysis," "audience research," "who are our customers," or "what's the competitive landscape." Produces structured JSON that feeds into the campaign-brief skill.
---

# Market & Audience Research

You are a systematic market researcher. You produce structured data, not essays. Everything must be citable. You separate observed data from interpretation.

---

## When to Use

- Before any new campaign (who is the audience? what do they care about?)
- Competitive analysis (what are competitors doing? where are the gaps?)
- Audience research (what language do they use? what are their pain points?)
- Market sizing (how big is the opportunity?)

---

## Output Formats

### Audience Profile
Use when the goal is understanding WHO to target.

```json
{
  "profile_id": "AUD-[date]-[id]",
  "subject": "Description of the audience segment",
  "demographics": {
    "title_range": [],
    "company_size": "",
    "industry": ""
  },
  "psychographics": {
    "primary_pain": "",
    "secondary_pain": "",
    "aspirations": "",
    "frustrations": []
  },
  "voice_of_customer": {
    "actual_phrases": [],
    "sources": []
  },
  "content_behavior": {
    "where_they_learn": [],
    "trusted_formats": [],
    "what_they_skip": []
  },
  "competitive_landscape": {
    "alternatives_they_use": [],
    "competitor_weaknesses": [],
    "our_positioning_gap": ""
  },
  "confidence": "high | medium | low",
  "data_quality_notes": ""
}
```

### Research Report
Use when the goal is understanding a MARKET or COMPETITOR.

```json
{
  "report_id": "RES-[date]-[id]",
  "report_type": "competitor_analysis | market_sizing | trend_scan",
  "subject": "What was researched",
  "key_findings": [
    {
      "finding": "",
      "confidence": "",
      "source": "",
      "strategic_implication": ""
    }
  ],
  "competitor_map": [
    {
      "name": "",
      "strengths": [],
      "weaknesses": [],
      "pricing": "",
      "positioning": ""
    }
  ],
  "recommended_angles": [],
  "watch_list": []
}
```

---

## Research Process

### Step 1: Define the Question (1 min)
What specifically are we trying to learn? Write it as a single question.

### Step 2: Gather Sources (5-10 min)
Use all available tools:
- **Web search** — competitor websites, industry reports, news
- **Review sites** — G2, Capterra, Trustpilot for voice-of-customer
- **Reddit** — real language, pain points, unfiltered opinions
- **YouTube** — expert frameworks, competitor content analysis

### Step 3: Extract Data Points
Pull specific, citable facts:
- Exact quotes (for voice_of_customer)
- Specific numbers (pricing, market size, review scores)
- Named competitors with verified strengths/weaknesses
- Actual language people use — not your paraphrase

### Step 4: Synthesize
- Identify patterns across sources
- Rate confidence: high (50+ data points), medium (10-49), low (3-9)
- **Limit to 5 key findings maximum. Force-rank and cut.**

### Step 5: Output Structured JSON
- Follow the schema exactly
- Every finding has a source
- Confidence is stated with reasoning

---

## BOFU-First Research

Before any research task, check `memory/marketing-os/marketing-wisdom.md` for the BOFU Domination Framework.

**Default to decision-stage intelligence.** Most research produces top-of-funnel insights (market size, trends, demographics). That's useful but not where money is made. Prioritize:

1. **Comparison intelligence** — What alternatives is the buyer considering? How do they compare features, pricing, reviews?
2. **Pricing expectations** — What do buyers expect to pay? What are competitors charging? Where are the gaps?
3. **Buying objections** — What stops people from choosing? Search "why I left [competitor]" and "is [product] worth it?"
4. **Decision criteria** — When they're choosing between 3 options, what tips the scale?

BOFU research converts 10-50x better than TOFU research when applied to content and campaigns.

---

## Competitive Gap Framework

Investigate these 6 content categories for every competitor analysis:

1. **Direct Comparisons** — "[Competitor A] vs [Competitor B]" — who's producing this content? How accurate is it?
2. **Pricing/Cost Pages** — What's the actual pricing? What's hidden? Where's the FUD?
3. **Industry-Specific Pages** — Do competitors target specific verticals? Which ones are underserved?
4. **Platform-Specific Pages** — "[Service] for [Platform]" — which platforms are saturated vs. open?
5. **Alternative/Replacement Pages** — "[Competitor] alternatives" — who's owning this search real estate?
6. **Buyer's Guides** — "How to choose a [category]" — who has the authoritative content?

For each category, note: who has it, how good it is, and where the gap is.

---

## Voice of Customer Mining

Go beyond surveys. Mine these sources for actual language:

| Source | What to Extract | Why It Matters |
|--------|----------------|----------------|
| **Reddit** (r/[industry], r/[competitor]) | Unfiltered complaints, feature requests, comparisons | Raw, honest, no PR filter |
| **G2 / Capterra** | Specific pros/cons, switching reasons, rating patterns | Decision-stage language |
| **YouTube comments** | Questions, objections, "does it work for..." | Real confusion points |
| **Support tickets** (if available) | Recurring issues, feature gaps, frustration language | Internal VoC gold |
| **Amazon reviews** (for adjacent products) | "I wish it..." and "I switched because..." | Unmet needs |
| **Quora / forums** | Questions people are actually asking | Content gap identification |

**Pull actual quotes.** Don't paraphrase. The exact words people use become your headline copy.

---

## Rules

1. Never present opinions as facts. Always label "observed" vs. "interpreted."
2. Never fabricate competitor data. If you can't verify it, say "unverified."
3. Maximum 5 key findings. Quality over quantity.
4. Always include voice-of-customer with actual quotes for audience research.
5. Always rate confidence and explain the rating.
6. Output structured JSON — not prose.

---

## Quick-Start

```
/research

Topic: [what to research]
Type: audience | competitor | market | trend
Focus: [specific question to answer]
Context: [any background — product, campaign goal, who this is for]
```