Scrape Trustpilot Reviews with DeepSeek, Analyze Sentiment with OpenAI

Workflow Name

Key Features and Highlights

This workflow automatically scrapes user reviews of specified companies from the Trustpilot website. It leverages the DeepSeek model to accurately extract key information from each review—such as author, rating, date, title, content, user country, and the total number of reviews posted by the user. Subsequently, it employs OpenAI to perform sentiment analysis, categorizing the review sentiment as positive, neutral, or negative. Finally, the structured review data and sentiment classification results are synchronized and updated in Google Sheets, enabling automated collection and intelligent analysis of review data.

Core Problems Addressed

Automates the collection of large volumes of Trustpilot user reviews, eliminating the tedious manual process of searching and copy-pasting reviews one by one.
Precisely extracts multidimensional information from reviews to ensure data completeness and accuracy.
Utilizes AI-driven sentiment analysis to quickly gain insights into customer feedback sentiment trends, helping businesses understand customer satisfaction and identify potential issues.
Enables real-time data updates and storage for convenient subsequent statistical analysis and business decision-making.

Application Scenarios

Brand Reputation Management: Automatically monitor and analyze customer reviews to promptly detect negative feedback.
Market Research: Gather user review data for competitors or own products to support product optimization.
Customer Service: Prioritize handling of negative reviews based on sentiment classification to improve customer satisfaction.
Data Reporting: Provide management with structured, dynamically updated customer feedback reports.

Main Workflow Steps

Manually trigger the workflow start.
Set the target company name and the maximum number of review pages to scrape.
Use HTTP requests to paginate and fetch review pages for the specified company on Trustpilot.
Extract the list of review links using HTML nodes.
Split and limit the number of review links as needed.
Fetch detailed HTML content for each individual review.
Invoke the DeepSeek model to extract information such as review author, rating, date, title, body, user country, and total review count.
Use the OpenAI model to perform sentiment analysis on the review body, classifying it as positive, neutral, or negative.
Query Google Sheets to check if the review already exists to avoid duplicate storage.
Append or update new review data and sentiment results into Google Sheets.

Involved Systems and Services

Trustpilot (review data source)
DeepSeek (information extraction AI model)
OpenAI (sentiment analysis AI model)
Google Sheets (data storage and management)
n8n Automation Platform (workflow orchestration)

Target Users and Value

Marketing professionals and brand managers: Easily grasp authentic customer feedback to optimize brand image.
Data analysts and product managers: Obtain structured user review data to support product decision-making.
Customer service teams: Quickly identify negative reviews to enhance customer experience.
Automation enthusiasts and developers: Implement complex scraping plus AI analysis automated workflows to save labor costs.

This workflow combines web scraping with advanced AI natural language processing technologies to efficiently and intelligently manage and analyze Trustpilot user reviews, making it a powerful tool for data-driven business operations.