Build an E-commerce Product Research AI Agent with n8n and Firecrawl

Input: a chat message (product + budget + optional region). Processing: AI plans search strategy → Firecrawl searches and scrapes review pages → AI analyzes and generates a report. Output: structured market analysis report with buying recommendations and seller opportunities.

Who

E-commerce sellers, dropshippers, affiliate marketers, or shoppers who need quick multi-marketplace product research and price comparison

You get

Generate a full market analysis report from a single chat request, including market overview, top 3 products, key insights, common complaints, and buy/sell recommendations

You need

n8n

Steps

1. Configure credentials and import the workflow

   Import this template workflow into n8n, then complete two credential setups: 1. Firecrawl API key: sign up free at firecrawl.dev and add the key to both Firecrawl nodes (🔍 Search Marketplaces and 📄 Scrape Product Pages). 2. LLM API key: add your OpenRouter or OpenAI API key to the Chat Model node (🤖 Gemini 2.5 Flash via OpenRouter). You may switch to another large-context model (e.g., GPT-4o) if preferred. Save and activate the workflow after configuration.

   Setup guidance · Editorial suggestion。Setup assistant starter; used to confirm credentials and model configuration after import

   Copy prompt

   Please help me verify the credential setup for this n8n workflow:
   - Is the Firecrawl API key filled in both the Search Marketplaces and Scrape Product Pages nodes?
   - Is the OpenRouter/OpenAI API key filled in the Chat Model node?
   - Is the selected model a large-context model (Gemini 2.5 Flash, GPT-4o, etc.)?
   List any missing credentials by node and type.

2. Send a research request

   Click "Open Chat" in the workflow and type a product research request. Include product name, budget, and optionally a region/marketplace. Examples (English): - "Research the best wireless earbuds under $30" - "Find the best robot vacuum on Amazon under $200" Examples (Arabic / Egyptian dialect): - "ابحثلي عن أحسن ماكينة قهوة في مصر أقل من 2500 جنيه" - "ابحثلي عن أحسن powerbank في السعودية أقل من 100 ريال" The Chat Trigger node receives the input and passes it to the AI Agent.

3. AI searches and scrapes product data

   The AI Agent (🧠 Product Research Agent) automatically plans a search strategy based on your input: - If a specific marketplace is mentioned (e.g., amazon.eg), search that site directly. - If only a country is mentioned (e.g., Egypt), search all major marketplaces for that country. - If no location is mentioned, search global review/comparison articles. The Agent calls Firecrawl Search (🔍 Search Marketplaces) to run the query, then calls Firecrawl Scrape (📄 Scrape Product Pages) to extract only the top 2 search results. These review pages typically contain data for 5–10 products, which is more efficient than scraping individual product listings.

   Execution prompt · Source text。🧠 Product Research Agent node systemMessage field (promptType=define)

   Copy prompt

   =You are a Product Research Agent specialized in e-commerce market analysis.
   
   When the user asks you to research a product:
   
   1. SEARCH: Use Firecrawl Search with a specific query like:
      "best [product name] under [price] [year] review"
      Example: "best wireless earbuds under $30 2025 review"
      DO NOT search directly on Amazon. Instead search for review articles and comparison pages that list multiple products with prices and ratings.
   
   2. SCRAPE: Scrape ONLY the top 2 results from the search. These review pages typically contain data about 5-10 products in one page, which is more efficient than scraping individual Amazon listings.
   
   3. ANALYZE: From the scraped review articles, extract product data and generate this report:
   
   📊 MARKET OVERVIEW
   - Number of products found
   - Price range (min / avg / max)
   - Average rating
   
   🏆 TOP 3 PRODUCTS
   - Product name, price, rating, key strength
   - Include purchase link if available
   
   💡 KEY INSIGHTS
   - Most common features customers want
   - What separates best-sellers from budget options
   
   ⚠️ COMMON COMPLAINTS
   - Top 3 recurring issues from reviews
   - Red flags to avoid
   
   ✅ RECOMMENDATION
   - Best product to BUY (best value)
   - Best niche opportunity to SELL (market gap)
   
   Rules:
   - Respond in the same language the user writes in
   - If user writes in Arabic, respond in Egyptian Arabic
   - Keep reports concise and actionable
   - Always include source URLs
   - If scrape data is insufficient, still provide analysis based on whatever data you found
   - NEVER say you cannot generate a report. Always provide the best analysis possible with available data.
   
   REGIONAL MARKETPLACE GUIDE:
   When the user mentions a region or country, use these marketplaces:
   
   Egypt (مصر):
   - amazon.eg (Amazon Egypt)
   - noon.com/egypt
   - jumia.com.eg
   
   Saudi Arabia (السعودية):
   - noon.com/saudi
   - amazon.sa
   - jarir.com
   - extra.com
   
   UAE (الإمارات):
   - noon.com/uae
   - amazon.ae
   
   Global / No region specified:
   - amazon.com
   - aliexpress.com
   - ebay.com
   
   SEARCH STRATEGY:
   - If user specifies a marketplace: search directly on that site
     Example query: "wireless earbuds site:amazon.eg"
   - If user specifies only a country: search on ALL marketplaces for that country
     Example query: "best wireless earbuds amazon.eg OR noon.com/egypt"
   - If user says nothing about location: search global review sites for comparison articles
     Example query: "best wireless earbuds under $30 2025 review"
   
   CURRENCY HANDLING:
   - If user mentions EGP/جنيه → prices in Egyptian Pounds
   - If user mentions SAR/ريال → prices in Saudi Riyals  
   - If user mentions AED/درهم → prices in UAE Dirhams
   - If no currency specified → prices in USD

4. Receive the structured market analysis report

   The AI Agent analyzes the scraped data and returns a structured report in the chat window, including: - 📊 Market Overview (number of products found, price range, average rating) - 🏆 Top 3 Products (name, price, rating, key strength, purchase link if available) - 💡 Key Insights (most wanted features, what separates best-sellers from budget options) - ⚠️ Common Complaints (top 3 recurring issues, red flags to avoid) - ✅ Recommendation (best product to buy + best niche opportunity to sell) The report is returned in the same language as the user input; if the user writes in Arabic, the response is in Egyptian Arabic.

Please help me verify the credential setup for this n8n workflow:
- Is the Firecrawl API key filled in both the Search Marketplaces and Scrape Product Pages nodes?
- Is the OpenRouter/OpenAI API key filled in the Chat Model node?
- Is the selected model a large-context model (Gemini 2.5 Flash, GPT-4o, etc.)?
List any missing credentials by node and type.