NoChannel Platform Ideas Portal
We need to access and export customer review data for the Furniture Row and Denver Mattress stores from the Upstart Commerce system in a format suitable for processing with Large Language Models (LLMs). This data will be used to create AI-generated summaries of customer reviews for each “store location”, enhancing our SEO efforts and providing valuable insights to potential customers, and generating new traffic to our sites.
Data Export Mechanism:
Develop a method to export all store location customer reviews from the Upstart Commerce system.
The export should be automated and scheduled to run regularly.
Data Format:
Export the data in a structured format, preferably JSON or CSV.
Each review should include, at minimum:
Review text
Rating
Date of review
Store ID or location identifier
Product information (if available)
Customer name (if available and permissible)
Store-Specific Segmentation:
Ensure that the exported data is segmented by store location.
Each store's reviews should be easily identifiable and separable from other stores' data.
API Utilization:
Upstart Commerce API, specifically the GET /reviews endpoint.
Update Mechanism:
Include a way to identify and export only new or updated reviews since the last export, to minimize data transfer and processing.
Documentation:
Provide clear documentation on how to access and use the exported data.
Include any necessary authentication or authorization details for accessing the data.
This exported review data will be used to generate AI-powered summaries for each store location. These summaries will be displayed on new, SEO-optimized pages on our website, providing potential customers with quick insights into each store's performance and reputation.
We aim to implement these page updates in Q4, so your prompt attention to this request is greatly appreciated.
If there are any limitations or potential issues with exporting this data from the Upstart Commerce system, please advise as soon as possible.
We're open to suggestions for optimizing this process or alternative approaches that might better suit our needs.
Thank you for your assistance in this project. Please let me know if you need any additional information or clarification.
Following our recent discussion regarding the feature, we have reviewed the requirements internally and have a few questions and clarifications:
Questions:
The PRD mentions including the customer's name, but with the condition "[if permissible]." Given that most compliance regulations restrict sharing user data, we have decided to exclude customer names from the review data. Is this acceptable?
The PRD does not specify the frequency of scheduled exports. Should the data be exported daily, weekly, fortnightly, or monthly? Please clarify the desired frequency.
Clarifications:
We plan to create a CRON job that will compile the data. The API endpoint will then provide a downloadable link for the compiled data.
The data format has been specified as either JSON or CSV. Each format has its advantages and challenges:
CSV may require additional cleanup due to columns with descriptions or textual data containing commas.
JSON, while comprehensive, may result in a large file with less human-readable data.
We suggest using Excel instead, where each sheet can represent data from one location (e.g., 100 sheets for 100 locations). Is this approach feasible, or should we proceed with either CSV or JSON?
We plan to develop this feature in a generic manner to ensure it can be deployed for product exports as well.
The scheduling feature will be made configurable with start and end dates to enhance its flexibility.
We look forward to your guidance on the above points.