| name | analyze |
| description | Collects market sizing, trend, and competitor data from the web and tags every claim with an evidence tier (primary > peer-reviewed/analyst > secondary press > anecdotal). Runs after `goal` has fixed the scope. Triggers on "시장 규모 알아봐", "경쟁사 조사", "이 시장 트렌드는?". |
Analyze — Evidence-Tiered Market & Competitor Research
Every number in a GTM deck traces back to a source someone can re-check.
This skill collects that evidence and tags it before anything gets
synthesized into a comparison or a plan.
Evidence Tiers (apply to every claim)
| Tier | Source type | Example |
|---|
| 1 — Primary | Company filings, official pricing pages, product docs, direct founder/exec statements, government stats | 10-K, official pricing page, census data |
| 2 — Analyst/peer-reviewed | Gartner/Forrester/CB Insights reports, academic studies, credentialed industry analysts | published market-sizing report |
| 3 — Secondary press | Reputable tech/business journalism citing sources | TechCrunch, Reuters covering a funding round |
| 4 — Anecdotal | Reddit/forum posts, unverified blog claims, single-user reviews, LinkedIn posts | "I heard their churn is high" |
A claim with no tier assigned is not usable. Tier 4 claims may be noted as
signal but must never anchor a number in the final deliverable.
Process
- Read
research/SCOPE.md (from goal). If missing, stop and invoke
goal first — do not guess the scope.
- Market sizing. Search for TAM/SAM/SOM figures, growth rate, and
adjacent-market comparisons. Prefer Tier 1–2 sources; cross-check any
single Tier 3 number against a second independent source before using it.
- Trend scan. Identify 3–5 structural trends (regulatory, technology,
buyer-behavior shifts) shaping the category over the stated horizon.
- Per-competitor collection (for each name in the seed list + any new
ones discovered): positioning/tagline, pricing (public tier or estimated),
target segment, key features, funding/scale signal, recent moves
(launches, pivots, layoffs), and 1–2 direct customer quotes/reviews if
available.
- Gap flagging. For any data point you could not find at Tier 1–3,
write "UNVERIFIED — tier 4 or absent" rather than silently omitting it or
guessing a number.
Write findings to research/FINDINGS.md:
---
scope_ref: research/SCOPE.md
created: {ISO-8601}
---
# Market & Competitor Findings
## Market Sizing
- TAM: {figure} — [Tier N, source, date](url)
- SAM: {figure} — [Tier N, source, date](url)
- Growth rate: {figure} — [Tier N, source, date](url)
## Trends
1. {trend} — [Tier N, source](url)
2. ...
## Competitors
### {Competitor name}
- Positioning: {...} — [Tier N](url)
- Pricing: {...} — [Tier N](url)
- Target segment: {...} — [Tier N](url)
- Key features: {...}
- Scale/funding signal: {...} — [Tier N](url)
- Recent moves: {...} — [Tier N](url)
- Customer signal: {quote or theme} — [Tier N](url)
(repeat per competitor)
## Unverified / Gaps
- {data point} — searched but not found at Tier 1–3; noted anecdotally only
Anti-Rationalization
- "This number sounds about right" is not a citation. Find the source or
mark it unverified.
- A single press mention repeated across three articles citing each other is
still one source, not three.
- Don't silently drop a competitor because data is thin — record the gap.
Evidence
Findings are complete when every market-sizing figure and every competitor
field has a tier + source link, or is explicitly flagged unverified.
Red Flags
- Numbers with no tier or link.
- Tier 4 sources (forum posts, anecdotes) used as the sole basis for a
market-sizing figure.
- A competitor section with all fields empty (thin research, not absence of
data) — flag for a second search pass instead of writing "unknown."