E-commerce sentiment analysis
Decoding Amazon beauty reviews.
I extracted and analyzed review text to understand which product features drew praise and which caused customer frustration.
The research scope
A star rating leaves a lot unsaid.
A four-star average gives a brand a quick read on performance. The review text explains why people felt that way. I built a lightweight tool in Google Sheets to explore 2,006 Amazon beauty reviews and compare star ratings with the sentiment and themes in the written feedback.
How I built it
Keep the analysis accessible in Sheets.
I processed the review dataset in Google Sheets, retaining timestamps, verified purchase status, helpful votes, and review text. I then categorized the text around the reasons customers gave for praise or frustration.
The analysis grouped feedback into user experience and application, product quality, and expectations versus reality. Native thematic cross-tabulations and sentiment distribution charts made those patterns easier to explore.
The dataset
2,006 reviews.
Review text alongside ratings, dates, purchase verification, and helpful votes.
The questions
What worked? What fell short?
The thematic breakdown separated comments about applying a product from comments about quality or unmet expectations. That gave the written feedback more context than a star average could provide on its own.