For the ManualTrigger Automate workflow, streamline the extraction and analysis of Trustpilot reviews. This automated process collects and organizes reviews, applies advanced clustering algorithms to identify key themes, and generates actionable insights. It efficiently exports results to Google Sheets, enabling businesses to understand customer sentiment and improve their services based on detailed feedback analysis.
This workflow is ideal for:
- Business Analysts looking to gain insights from customer reviews.
- Product Managers wanting to understand customer feedback and improve products.
- Marketing Teams focused on enhancing brand reputation through customer sentiment analysis.
- Data Scientists interested in clustering algorithms and natural language processing.
- Small Business Owners who want to leverage customer feedback to enhance their services.
This workflow addresses the challenge of extracting valuable insights from Trustpilot reviews. It automates the process of scraping reviews, analyzing sentiment, and clustering similar feedback. By utilizing advanced algorithms, it helps in identifying common themes and suggestions from customers, enabling businesses to make informed decisions based on real customer experiences.
2. Fetch Trustpilot Pages: It retrieves the latest reviews from Trustpilot for the specified company, scraping up to 3 pages of reviews.
3. Extract Reviews: The HTML content is processed to extract relevant data such as review author, rating, title, text, and dates.
4. Prepare Data for Vector Storage: The extracted reviews are formatted and stored in a structured format suitable for further analysis.
5. Store Reviews in Qdrant: The reviews are vectorized and inserted into the Qdrant vector store, allowing for efficient similarity searches and clustering.
6. Find Reviews: The workflow retrieves stored reviews based on specific criteria, including date ranges and company ID.
7. Apply K-means Clustering: A K-means clustering algorithm is applied to group similar reviews, facilitating the identification of common themes.
8. Generate Insights: The clustered reviews are analyzed using a language model to produce insights, sentiments, and suggested improvements.
9. Export to Google Sheets: Finally, the insights and raw responses are exported to a Google Sheet for easy access and reporting.
Users can customize this workflow by:
- Modifying the Company ID: Change the companyId variable in the Set Variables node to target a different company.
- Adjusting the Review Fetch Limit: Modify the pagination settings in the Get TrustPilot Page node to scrape more or fewer pages of reviews.
- Tuning Clustering Parameters: Adjust the parameters in the Apply K-means Clustering Algorithm node to change the number of clusters or clustering logic.
- Altering Insight Generation Prompts: Customize the prompts in the Customer Insights Agent node to tailor the type of insights generated from the reviews.
- Changing Export Settings: Modify the Export To Sheets node to change the target Google Sheets document or customize the columns being populated.