Competition Analysis

Review Mining -> Opportunity Matrix (Comet Prompt)
Inputs (fill these)
PRODUCT_NAME: {your product name}
CATEGORY: {e.g., leggings, manual coffee grinder, e-skateboard}
REGION/MARKET: {US | EU | Global}
DATE_RANGE: {last 6 months | last 12 months}
CHANNELS (comma-separated): {Amazon, Walmart, brand PDPs, Reddit r/, YouTube reviews, TikTok/IG posts, specialty retailer listings}
KEYWORDS to bias discovery (optional): {fit, durability, shipping damage, battery, returns, sizing, app UX}
COMPETITORS to include (optional): {comma-separated list}
EXCLUSIONS (optional): {exclude marketplaces/brands/categories}
CURRENCY: {USD | EUR | …}
Mission
Mine recent public reviews and discussion to surface where competitors fail and buyers feel underserved. Convert the findings into:

a prioritized product/ops roadmap,
a crisp Opportunity Matrix (issues → fixes → differentiation claims), and
immediately deployable FAQ/macros + PDP/ad copy.
What to Collect
From each relevant source in {CHANNELS} within {DATE_RANGE}:

Item: product/competitor name, URL, source, date
Metadata: star rating (if any), view count (if video), comment/reply count
Extracted fields:
Topic/feature (normalize to a controlled list)
Sentiment (pos/neg/neutral)
Issue pattern (short phrase)
Root-cause hints
Evidence snippet (≤25 words, verbatim, quoted)
Price anchor (if visible) in {CURRENCY}
Access date
Aim for 300–600 total data points when possible (or exhaust sources).
Processing & Analysis
Normalize & de-duplicate brands/products; remove marketplace duplicates.
Theme clustering by semantic similarity; label clusters clearly.
Scoring per cluster:
Frequency (count)
Severity (1–5)
Revenue Impact (Low/Med/High)
Momentum (trend over time)
Prioritize by Frequency × Severity × Impact (break ties with Momentum).
Opportunity Matrix (top 8–15 themes). For each theme:
Evidence summary (2–3 linked quotes)
Proposed Fix (product/packaging/policy/app copy)
Differentiation Claim (defensible, customer-language)
Effort (S/M/L), Owner (Product, Ops, CS, Marketing), Expected Impact (Rev/Returns)
Test Ideas (PDP copy, ad hook, FAQ change, UX tweak)
Roadmap: Top 10 fixes with owner, effort, impact, and next step.
FAQ & Macros:
5–10 new/updated FAQ entries (PDP + Help Center)
6–10 CS macros (pre-purchase + post-purchase)
Copy Kit:
3 PDP headline + subhead options
3 ad hooks with 3-beat structures
3 email/SMS snippets
Output Format (in this order)
Theme Cluster Table (Markdown)
Columns: Rank · Theme · Mentions · Severity(1–5) · Revenue Impact · Momentum · Key Evidence (short quotes + links)
Opportunity Matrix (Markdown)
Columns: Theme · Evidence (links/quotes) · Proposed Fix · Differentiation Claim · Effort · Owner · Expected Impact · Test Ideas
Prioritized Roadmap (Markdown list) — Top 10 fixes with owner/effort/impact/next step
FAQ and CS Macros (copy-paste blocks)
Copy Kit (PDP headlines, ad hooks, email/SMS snippets)
Downloads
themes.csv — raw clustered items
opportunity_matrix.csv — matrix rows
evidence.jsonl — per-item extracted records with source metadata
JSON Summary (code block named review_mining_summary):
{

"product": "{PRODUCT_NAME}",

"category": "{CATEGORY}",

"region": "{REGION/MARKET}",

"date_range": "{DATE_RANGE}",

"channels": ["..."],

"themes_top": [

],

"top_claims": [

],

"roadmap_top": [

],

"files": {

"themes_csv":"themes.csv",

"matrix_csv":"opportunity_matrix.csv",

"evidence_jsonl":"evidence.jsonl"

},

"sources_count": 0

}

Selection Rules
Focus on close substitutes in {CATEGORY} and {REGION/MARKET}; include {COMPETITORS} if provided.
Exclude {EXCLUSIONS}.
Prefer primary listings/pages and widely viewed reviews; use forums for qualitative depth.
Evidence & Citation Rules
Each theme must include at least two source URLs and ≤25-word quotes with access dates.
Prefer newest dated sources; when conflicting, show both values and a brief reconciliation note.
If price/specs are not visible, mark unknown or contact sales.
Robust Retry & Resilience
Access failures (404/500/timeouts/CAPTCHAs/paywalls): retry up to 2 times with backoff (1s, then 3s). If blocked, use alternate authoritative pages (pricing/FAQ/docs/cached). Mark paywalled:true and continue.
Rate limits/tool errors: rephrase queries; pivot to site:help, site:docs, site:pricing, or public review aggregators. Limit concurrent fetches to 5 pages.
De-duplication: normalize brand/product URLs; drop near-dupes.
Quality gate: don’t include a top theme without ≥5 independent mentions unless severity is critical.
Deliverable Style
Crisp bullets, plain language.
Use {CURRENCY} for prices.
Neutral tone; put recommendations in the Opportunity Matrix and Roadmap.
End with “What to test this week” listing the top 3 claims and exact PDP/ad placements.
Now Run
Discover sources in {CHANNELS} for {CATEGORY}/{PRODUCT_NAME} within {DATE_RANGE}.
Extract and normalize data as specified.
Cluster, score, and rank themes.
Build the Opportunity Matrix, Roadmap, FAQ/macros, and Copy Kit.
Export the CSV/JSON files and present outputs in the order above with live citations.

https://docs.google.com/document/u/0/d/1TxVIAibn5aRU9FbnAy5imdKRhNX4Z5uwjioiN58QTEg/