asin-deep-research

name	asin-deep-research
description	One-shot ASIN research package using SellerApp via n8n MCP. Pulls product details, reviews (1-5★ with sentiment), Rufus queries, reverse ASIN keywords, competitor SERP, LQI score, and BSR history into a pre-filled CRO Research Brief. Use when starting any new ASIN engagement, before content planning, or when running `asin-deep-research {ASIN}`. Replaces 3-4 hours of manual data pulling.

asin-deep-research — Foundational ASIN Research Pipeline

Single command that produces a complete Research Brief (Sections 1-5 of MASTER-CRO-REFERENCE.md §2) for any ASIN. This is the foundation skill — every CRO project starts here.

Invocation

asin-deep-research {ASIN}
asin-deep-research {ASIN} {marketplace}

• ASIN (required): Amazon product ID (e.g. B0CXG3HMX1)
• marketplace (optional, default: us): us, uk, de, ca, etc.

Output

Writes to /tmp/cro-research/{ASIN}-research-brief-{date}.md and opens with open command.

Tools Used

All SellerApp data flows through the n8n MCP wrapper (workflow 9RmjDT107uXtrImf at https://keplo.app.n8n.cloud/workflow/9RmjDT107uXtrImf). Tool prefix: mcp__f32016b6-7c77-45e2-b4a2-70195c5f2d2d__*.

Step	n8n MCP tool	Purpose
Product baseline	Get_Product_Details (with all flags)	ASIN, title, brand, BSR, price, ratings, sales estimate
Review mining	Get_Product_Reviews (3 pages each at 1★, 3★, 5★, sort=recent)	Purchase drivers / objections / surprises / questions / photo refs
Rufus signal	Get_Rufus_AI_Queries	Customer questions Amazon's AI surfaces
Keyword landscape	Keyword_Research_V2_Reverse_ASIN (results_count=200)	High-volume search terms + benefit signals
Competitor SERP	Keyword_Search_Result_SERP_ (extended_response=1) on top 3 keywords	Top-10 competitors per primary keyword
Quality baseline	LQI request (via Get_Product_Details LQI flag if available, else flag as manual)	Section-level quality score
BSR trajectory	Get_Product_History_30d_	30-day price/BSR/rating trend

Phase 1 — Collect (parallel where possible)

Run these in a single batch where the MCP supports it:

1. Get_Product_Details for ASIN with full flags (product_specifications=1, ratings=1, price_detail=1, potential_detail=1, promotions=1)
2. Get_Product_Reviews page 1, sort=recent (latest 10 reviews — recency baseline)
3. Get_Rufus_AI_Queries for ASIN
4. Keyword_Research_V2_Reverse_ASIN for ASIN with results_count=200
5. Get_Product_History_30d_ for ASIN (price + BSR + rating + review_count)

Then sequentially: 6. Get_Product_Reviews pages 1-3 at rating=1 (objections) 7. Get_Product_Reviews pages 1-3 at rating=5 (drivers) 8. Get_Product_Reviews page 1 at rating=3 (nuanced) 9. From step 4, identify the top-3 keywords by relative_score → Keyword_Search_Result_SERP_ extended for each

Phase 2 — Synthesize into Research Brief

Map the data into the 5 brief questions from ~/.claude/knowledge/cro-methodology/decision-framework.md and MASTER-CRO-REFERENCE.md §2:

Section 1 — Customer Profile

• Demographics from photo reviews + targeted keywords (e.g. "for small apartments")
• Use context (when/where/why)
• Pull source: review images, lifestyle keywords

Section 2 — Why They Buy (Top 3-5 Purchase Drivers, ranked)

• 5★ review themes ranked by frequency
• High-volume benefit keywords from reverse ASIN
• Tag each driver: frequency >30% = mandatory visual; frequency 10-30% = supporting

Section 3 — Worries / Objections (Top 3-5)

• 1-3★ review themes ranked by frequency
• Rufus "concern"-classified queries
• Recurring questions across all stars
• Tag each: >20% of negative reviews = preempt with image

Section 4 — Competitive Landscape

• Top 10 competitors per primary keyword (from SERP pulls)
• Table stakes (every competitor shows X)
• Quality benchmarks (BSR / rating / image count averages)
• Differentiation gaps (no one shows X but customers care)

Section 5 — Visual Must-Show List

• Cross-reference: every Section-2 driver → "is this visually demonstrated?"
• Every Section-3 worry → "is this preempted?"
• Output: prioritized image-slot assignments per 02-visual-content/listing-images.md

Phase 3 — Output

Write to /tmp/cro-research/{ASIN}-research-brief-{YYYY-MM-DD}.md. Required header:

# Research Brief — {Title}

**ASIN:** {ASIN} | **Brand:** {Brand} | **Marketplace:** {geo} | **Date:** {date}

## Data Sources

| Source | Records | Status |
|---|---|---|
| Product Details | 1 ASIN | ✅ |
| Reviews | {N} reviews across 1★/3★/5★ | ✅ |
| Rufus Queries | {N} queries | ✅ / ⚠️ Empty |
| Reverse ASIN Keywords | {N} keywords | ✅ |
| Competitor SERP | Top-{N} on {K} keywords | ✅ |
| BSR / Price History | 30 days | ✅ |
| LQI | {Score} / Manual | ✅ / ⚠️ Manual needed |

## Quick Stats

- BSR: #{rank} in {category} ({trend over 30d})
- Rating: {n}★ ({count} ratings)
- Price: ${price} ({trend})
- Sales estimate: {low}–{high} units/day
- Number of sellers: {N}
- Variations: {parent or N variants}

Then sections 1-5 as above, ending with:

## Recommended Next Skills

- For SERP differentiation: `competitor-sweep {ASIN}` (deep version of competitor data)
- For review drill-down: `review-mining {ASIN}`
- For main image work: `main-image-pipeline {ASIN}`
- For full content plan: `/cro-content-plan {ASIN}`

Reference Files

• ~/.claude/skills/cro/methodology.md — full diagnostic framework
• ~/.claude/skills/cro/review-analysis.md — 8-layer claim decomposition
• ~/.claude/skills/cro/sellerapp-api.md — endpoint reference (legacy direct API; this skill uses the n8n MCP wrapper instead)
• ~/.claude/knowledge/cro-methodology/decision-framework.md — 5-check sequence
• Vault: CRO-Knowledge-Base/MASTER-CRO-REFERENCE.md §2 (The Research Brief)
• Vault: CRO-Knowledge-Base/01-research/building-research-brief.md

Quality Bar

Before declaring "research brief ready":

• Every section answers the question (no "TBD" or empty bullets)
• Every claim cites the data source ("from 1★ reviews" / "from Rufus" / "from keyword X with vol N")
• Frequency tags applied to drivers and worries
• Section 5 has at least one visual must-show per slot 1-7
• Data Sources header lists every source with status
• Output file exists at /tmp/cro-research/ and was opened

Failure Modes

Symptom	Likely cause	Action
Rufus returns empty	Newer ASIN, low traffic	Note in Data Sources, proceed without; flag for manual SERP browse
Reverse ASIN returns < 20 keywords	Low-traffic ASIN	Pull Keyword_Search_Result_SERP_ for the brand/category as fallback
Reviews < 50	Newer ASIN	Pull all available, prioritize photo reviews + verified
MCP timeout	n8n workflow stalled	Retry once, then check workflow 9RmjDT107uXtrImf health

Auto-Triggers

This skill runs automatically (without asin-deep-research prefix) when:

• User pastes an Amazon ASIN or amazon.com/dp/ link as the start of a new CRO conversation
• User asks "research this ASIN" / "pull data on B0..."
• A higher-level pipeline (/cro-content-plan, main-image-pipeline) needs the brief