Challenges
Our client is an e-commerce affiliate, USA listed on Amazon. The project dealt with Content Creation and Content Upload. The client wanted to build their e-commerce affiliate business from scratch and for that, they needed engaging content, minimize manual intervention and analytics using machine learning. Automation included all parts of chain-like product review scrapping, review analysis, image upload, product descriptions, product ranks. Moreover, the client did not want to use a third-party analytics tool.
Solution
Since our client was very particular about what they wanted, we used their requirements as a starting point and gave them solutions that fit their needs. For example, we used Selenium, ML to provide automation, and analytics. A flexible website was developed on WordPress from scratch.
Automating the entire data pipeline was a technical challenge. Automation included product description, product reviews analysis, image from s3 uploaded to the website, automation of QA reports. The process included verifying that the image actually belonged to the said product and delinking them from the website including a crawler to report incorrect images and text in real-time. We also created engaging content which is always a challenge in the competitive and crowded e-commerce market. Whilst the basics were covered for each item, two new information pieces were added which were product features and review analytics. The product description incorporated most talked about features of the product highlighting the key aspect of each item. The product review analysis included sentiment analysis modeling using ML.
Outcome
More than 2000 engaging product pages were created for the brand. It now takes less than 30 minutes to create a new product page without manual intervention. Using their own data, ML models were trained and developed for feature extraction, sentiment analysis with high accuracy.