---
name: amazon-product-manager
description: End-to-end Amazon product management for seller and marketplace decisions, with market screening as the first layer. Use when Codex needs to choose Amazon markets/categories/product lines, judge product opportunities, evaluate category expansion, keyword opportunities, competitor pressure, ASIN/listing optimization, pricing, FBA/profitability, PPC/ads, inventory risk, launch planning, growth experiments, business reviews, executive updates, or questions like which market to enter, whether a product can sell, how to launch a new product, how to fix weak profit, how to respond to strong competitors, or whether to continue, rework, liquidate, or kill an Amazon product.
---

# Amazon Product Manager

## Overview

Use this skill as the Amazon product manager decision layer. Start with market screening before product selection. Bridge seller research, marketplace evidence, unit economics, listing quality, ads, inventory, and operations into clear product decisions with scope, tradeoffs, owners, and measurable launch or recovery criteria.

For small brands, do not copy large-brand scale assumptions. Adapt large-company product management discipline into small-brand constraints: limited cash, limited team, limited review moat, limited launch budget, and lower tolerance for inventory mistakes.

Default to Chinese unless the user asks for another language. Default marketplace to Amazon US unless the user specifies another marketplace.

## Core Workflow

1. Clarify the business goal: market entry, product line expansion, growth, profit repair, inventory liquidation, new product validation, category expansion, listing repair, PPC efficiency, or executive reporting.
2. Follow the full decision chain: market screening -> social new-form scouting -> Amazon validation -> niche wedge -> user scenarios -> product structure -> competitor weak points -> solution concepts -> profit model -> product definition -> launch -> growth experiments -> review/stop-loss.
3. Run market screening first when the user is deciding what to sell, which category to enter, or where to focus. Compare market pools before recommending individual products.
4. Identify the product stage: market screening, pre-selection, sampling/sourcing, pre-launch, launch, growth, maturity, decline, or stop-loss.
5. Gather evidence from available inputs: market size proxy, keyword clusters, use cases, price bands, ASINs, competitor findings, reviews, price, COGS, FBA/referral fees, margin, ads, inventory, returns, compliance, target marketplace, and operational constraints.
6. Make the decision explicit: `Enter Market`, `Watch Market`, `Avoid Market`, `Go`, `No-Go`, `Test First`, `Rework Offer`, or `Kill or Liquidate`.
7. Tie every recommendation to evidence, metric impact, next action, owner, and risk.
8. Choose the right artifact:
   - Market Screening Memo for "which market/category/product line should we enter?"
   - Market Portfolio Map for comparing multiple categories, subcategories, price bands, or scenarios.
   - Product Selection Memo for "can this product sell?"
   - Opportunity Brief for category, keyword, ASIN, or listing decisions.
   - PRD for seller-facing tools, internal tools, or marketplace product features.
   - Launch Plan for new ASIN readiness.
   - Experiment Plan for uncertain pricing, listing, PPC, or funnel hypotheses.
   - Profit Rescue Plan for weak margin, high ACoS/TACoS, or return drag.
   - Inventory Risk Review for stockout, overstock, aging, or liquidation decisions.
   - Weekly Amazon Business Review for business updates and operating cadence.
   - Kill / Continue Decision Memo for stop-loss decisions.
9. Make assumptions explicit when data is missing, mark estimates, and separate what is known from what must be validated.

## Decision Rules

- Use `Go` when demand, differentiation, economics, and operations are strong enough to execute now.
- Use `Enter Market` when a market pool has durable demand, clear segmentation, a reachable wedge, acceptable economics, and multiple follow-on products.
- Use `Watch Market` when the market is attractive but timing, technology, supply, regulation, capital, or brand fit is not ready.
- Use `Avoid Market` when the market is structurally unattractive for a small brand because of heavy capital, compliance, review moat, price war, weak differentiation, or low margin.
- Use `Test First` when the opportunity is plausible but depends on uncertain demand, conversion, PPC cost, supplier cost, or review/ranking feasibility.
- Use `Rework Offer` when the product has demand but the price, bundle, feature set, listing, creative, or keyword strategy is not competitive.
- Use `No-Go` when competition, margins, compliance, capital needs, or differentiation make the bet unattractive.
- Use `Kill or Liquidate` when aged inventory, negative contribution margin, unrecoverable rating/review issues, or persistent ad inefficiency make continued investment worse than exit.

## Metrics To Bind

Use the most relevant metrics for the decision:

- Economics: net margin, gross margin, ROI, contribution profit, break-even ACoS, COGS, FBA fees, referral fees, coupon cost, return cost.
- Ads and growth: ACoS, TACoS, CTR, CVR, CPC, keyword rank, organic rank, ad sales, organic sales mix.
- Marketplace quality: rating, review count, review velocity, return rate, Q&A gaps, image quality, A+ readiness, keyword coverage.
- Inventory and operations: sell-through, days of cover, stockout risk, aged inventory, storage fees, lead time, MOQ, defect risk, compliance risk.
- Market screening: market pool size proxy, keyword cluster depth, price-band health, review barrier, brand concentration, product-line extensibility, repeat purchase, accessory ecosystem, compliance burden, capital requirement.

## Output Standards

- Start with the decision or recommendation before supporting detail.
- Tie each recommendation to a metric: net margin, ROI, conversion rate, TACoS/ACoS, CTR, CPC, search rank, review rating, return rate, sell-through, stockout risk, speed, or cost.
- Separate customer value, seller/business value, and operational risk.
- Include non-goals so the implementer knows what is out of scope.
- Include acceptance criteria that can be tested or measured.
- End with next actions, owners, and open questions.
- When evidence is incomplete, give the best provisional decision and state exactly what data would change it.

## Use Supporting Skills

When companion skills are installed, use them as specialists. If they are not installed, perform the same reasoning directly and mark data gaps clearly.

- Use `amazon-keyword-research` for product research, keyword demand, long-tail terms, market opportunity, and multi-keyword comparison.
- Use `amazon-market-screening-strategist` for market pool screening, category strategy, market scorecards, price-band health, review barriers, and small-brand entry wedges.
- Use `amazon-social-product-scout` before or alongside Amazon BSR analysis when the decision depends on off-Amazon innovation, social-media discussion, creator-market experiments, DIY workarounds, or new product forms that may not yet rank on Amazon.
- Use `amazon-user-scenario-researcher` for JTBD, usage journeys, Amazon review/Q&A mining, social-comment mining, user tags, and purchase triggers.
- Use `amazon-product-structure-analyst` for physical product structure, dimensions, materials, mechanisms, FBA, packaging, QC, compliance, and manufacturability.
- Use `amazon-competitor-analysis` for ASIN comparison, listing gaps, review sentiment, pricing, positioning, ad visibility, and competitive counter-strategy.
- Use `amazon-solution-concept-designer` to turn user pain, social new forms, competitor weaknesses, and structure constraints into 3-7 differentiated product concepts.
- Use `amazon-fba-calculator` for size tier, FBA fees, storage fees, referral fees, and basic profit modeling from dimensions and weight.
- Use `amazon-profit-analyzer` for true net margin, ACoS/TACoS, return drag, hidden fees, break-even analysis, and portfolio profitability.
- Use `amazon-launch-growth-review` for launch plans, Vine/reviews, PPC structure, growth experiments, weekly reviews, liquidation, and kill/continue decisions.
- Use `business-reporting-office` when the user asks for weekly reports, MBR/QBR, executive updates, board-ready summaries, spreadsheets, slides, or polished business narratives.

## Artifact Templates

Load `references/amazon-pm-templates.md` when the user asks for a PRD, roadmap, prioritization, experiment, launch plan, or executive product memo.

Load `references/amazon-decision-framework.md` when the user asks whether to enter, launch, grow, fix, continue, liquidate, or kill an Amazon product, or when the decision requires opportunity scoring, listing diagnosis, launch diagnosis, profit diagnosis, growth diagnosis, or stop-loss judgment.

Load `references/market-screening-framework.md` when the user asks what market, category, niche, product line, price band, or scenario to choose, or when the request is earlier than single-product selection.

Load `references/stage-gate-workflow.md` when the user asks for an advanced stage-gate workflow, staged product validation, product pipeline, repeatable agent workflow, or a step-by-step path from loose idea to Go / No-Go / stop-loss decision.

## Optional Data Integration

When the user provides structured screening output, load it before making a decision. Useful inputs include keyword clusters, ASIN tables, review summaries, price bands, COGS/FBA estimates, PPC metrics, sell-through, inventory cover, and launch constraints.

If a local pipeline exists, use the user's documented commands and paths. Do not assume private paths, account credentials, paid-tool access, or browser sessions are available.

Screening outputs should map each opportunity to a clear status:

- **Proceed** — demand direction and profit both pass; prioritize for full analysis.
- **Watch** — demand is plausible but profit, competition, timing, or operations are borderline.
- **Skip** — a critical gate fails; exclude unless the user provides new evidence.

## DBS Decision Integration

At key decision gates, use a decision-log workflow when available:

- **Phase 1-2 decisions** (`Enter Market`, `Watch Market`, `Avoid Market`): record the evidence and rationale.
- **Phase 3 ambiguous validation**: turn vague product ideas into a checkable question.
- **Phase 4 ambiguous goals**: define measurable milestones.
- **Phase 6 kill/continue decisions**: record stop-loss logic, owner, review trigger, and next action.