Adds dataset checks, reproducible counts, traceable quotations, and sampling limits.
Analyze the customer-review file or review text supplied for [COMPANY/PRODUCT]. Use only these reviews.
Inspect the columns, date range, missing values, and duplicates. Report the actual number of usable reviews rather than assuming [N/AMOUNT] is correct. Keep stable review IDs; do not expose unnecessary personal data.
Code recurring positive and negative themes. Count distinct reviews mentioning each theme, report the denominator and percentage, and allow a review to have multiple themes. Use a data-analysis tool for counting when available; say if results are a manual sample.
Return a pros/cons table with theme, count, percentage, a short exact quotation with review ID, and a practical implication. Flag contradictory feedback, small samples, and selection bias. Distinguish customer statements from your interpretation; never invent quotations or extrapolate to all customers.