User Review Analysis & Summarization

1. Overview

This process takes a collection of customer reviews for a single product, extracts the main positive points (pros), the main negative points (cons), and the overall sentiment expressed by shoppers. The result is a concise, easy‑to‑read summary that highlights the most common feedback themes.

2. Business Value

Quick Insight: Provides product managers with a rapid snapshot of what customers love and dislike, without reading every single review.
Data‑Driven Decisions: Helps prioritize product improvements, marketing messages, and feature road‑maps based on real user sentiment.
Time Savings: Saves hours of manual review and ensures consistent, unbiased summarization.

3. Operational Context

When to run:
- When preparing a product performance review, a monthly product health check, or before a product launch decision.
Who uses it:
- Product managers, product owners, and product‑marketing teams in e‑commerce organizations.
How often:
- Typically after a set period (e.g., weekly, monthly) or whenever a new batch of reviews is available.

4. Inputs

4.1 Customer Review List

Provide a list of individual reviews for a single product. Each review must contain the fields listed below. The list can be supplied as a simple array (e.g., in a spreadsheet or JSON‑style list) – the system will treat it as a collection of items.

Field	Description	Example
Product Name	The name of the product being reviewed	“BlueWater Blender”
Review Date	Date the review was posted (YYYY‑MM‑DD)	“2025‑07‑12”
Reviewer Name (optional)	Name of the reviewer or “Anonymous”	“John D.”
Rating	Customer rating on a 1‑to‑5 scale, where 1 = very unhappy, 5 = very happy	4
Review Text	Full text of the customer’s comment	“The motor is powerful but the blender is noisy.”

Note: Every review must have at least a Product Name, Rating, and Review Text. If any required field is missing, the process will flag that review for manual review.

5. Outputs

5.1 Review Summary

A plain‑language summary containing the following sections, presented as bullet points unless otherwise noted.

Output Name	Contents	Formatting Rules
Product Name	Copied from input	Title‑case
Summary Date	Date the summary was generated (YYYY‑MM‑DD)	ISO format
Overall Sentiment	One of: “Positive”, “Neutral”, “Negative”	Capitalized
Average Rating	Average of all numeric ratings, rounded to one decimal place	Numeric (e.g., 4.2)
Pros Summary	List of the most‑common positive points (max 5 bullet items)	Bullet points, no duplication
Cons Summary	List of the most‑common negative points (max 5 bullet items)	Bullet points, no duplication
Key Positive Sentiments	Short statements (max 5) that capture the strongest positive feedback	Sentence case
Key Negative Sentiments	Short statements (max 5) that capture the strongest negative feedback	Sentence case
Sentiment Score	Simple classification based on average rating: 4–5 = “Positive”, 2–3 = “Neutral”, 1 = “Negative”	Word (Positive/Negative/Neutral)

All sections should be presented in plain text, with each list item on a new line and preceded by a hyphen (“-”) for bullet points.

6. Detailed Plan & Execution Steps

Collect Reviews – Gather all review items for the target product into the “Customer Review List”.
Validate Input – Verify that each item contains Product Name, Rating, and Review Text.
- If any required field is missing, mark the review as “Missing data – review manually”. Continue with the remaining items.
Extract Positive Statements –
- Scan each review text.
- Capture any phrase that expresses a positive experience (e.g., “fast shipping”, “great quality”).
- Store each positive phrase in a temporary list.
Extract Negative Statements –
- Scan each review text.
- Capture any phrase that expresses a negative experience (e.g., “broken”, “slow”).
- Store each negative phrase in a temporary list.
Classify Sentiment per Review –
- Use the Rating: 4‑5 → Positive, 2‑3 → Neutral, 1 → Negative.
- Tag each review accordingly.
Aggregate Pros & Cons –
- Count how many times each phrase appears (case‑insensitive).
- Sort by frequency and keep the top five unique items for each list.
- If fewer than five unique items exist, list all that are available.
Generate Sentiment Summary –
- Compute Average Rating (average of all numeric ratings).
- Translate the average rating to Overall Sentiment based on the classification in step 5.
Craft the Review Summary –
- Write a one‑sentence Product Name line.
- Add Summary Date (today’s date).
- Add Overall Sentiment and Average Rating.
- List Pros Summary, Cons Summary, Key Positive Sentiments, and Key Negative Sentiments as separate bullet‑point sections.
- Add the Sentiment Score (same as Overall Sentiment).
Finalize Output – Present the complete Review Summary according to the formatting rules in the “Outputs” table.

7. Validation & Quality Checks

Mandatory Fields – Confirm each processed review had a rating and text. Missing fields cause a “manual review” flag.
Average Rating Accuracy – Re‑calculate the average rating after removal of any flagged reviews; verify it matches the reported value.
No Duplicate List Items – Ensure the lists of pros, cons, and sentiment points contain no duplicate statements.
Sentiment Alignment – Verify that Overall Sentiment matches the Sentiment Score derived from the average rating.
Complete Sections – Ensure every required output section appears. If any section is missing, flag the entire output as “Incomplete – check input”.

8. Special Rules / Edge Cases

No Cons Found – If no negative statements are identified, list “No significant cons identified.”
No Pros Found – If no positive statements are found, list “No notable positives identified.”
Small Sample Size – If the total number of reviews is fewer than 5, add a note: “Sample size small – interpret findings cautiously.”
All Reviews Same Rating – Still calculate average and proceed; the sentiment classification will be consistent.
Non‑English Review – If a review is not in English, skip it and add a “Non‑English review omitted” note.
Missing Rating – Skip the review and add a “Rating missing – manual review required” flag.

9. Example

Input (Customer Review List)

Product Name: “BlueWater Blender”
Review Date: 2025‑07‑02 – Rating: 5 – Review Text: “The motor is incredibly powerful and makes smoothies in seconds.”
Review Date: 2025‑07‑04 – Rating: 4 – Review Text: “Easy to clean, but the blender is a little loud.”
Review Date: 2025‑07‑10 – Rating: 2 – Review Text: “The blender works but the lid doesn’t fit well, causing spills.”

Output (Review Summary)

Product Name: BlueWater Blender
Summary Date: 2025‑08‑11
Overall Sentiment: Positive
Average Rating: 3.7

Pros Summary

Powerful motor makes quick smoothies.
Easy to clean.

Cons Summary

Loud operation.
Lid does not fit securely.

Key Positive Sentiments

“Powerful motor.”
“Easy to clean.”

Key Negative Sentiments

“Loud.”
“Lid doesn’t fit, causing spills.”

Sentiment Score: Positive

Appendix A – FAQ

What if a review contains both a pro and a con? Both statements are captured: the pro goes into the “Pros Summary” and the con into the “Cons Summary”.
How many reviews are needed for a reliable summary? The more reviews, the better. A minimum of 5 is recommended; fewer than 5 will still be processed but flagged as a small sample.
What if a review does not have a rating? The review is excluded and flagged for manual follow‑up.
Can the process handle multiple products at once? This SOP is for one product per run. To handle multiple products, run the process separately for each product.
Are the summary bullet points limited in length? Keep each bullet short – one sentence or phrase.
How are ties in frequency resolved? If two statements appear the same number of times, keep the first encountered in the original list.
What if the same phrase appears with different wording? Phrases are considered the same if they share the core meaning after removing minor word differences (e.g., “fast shipping” vs. “quick shipping” are merged).
Is the summary suitable for marketing copy? Yes – the “Pros Summary” can be repurposed for marketing, but verify compliance with brand guidelines.
What does “Overall Sentiment” mean? It is a high‑level sentiment (Positive, Neutral, Negative) derived from the average rating.

Appendix B – Glossary

Term	Definition
Pros	Positive aspects or benefits mentioned by customers.
Cons	Negative aspects or issues mentioned by customers.
Sentiment	The overall feeling expressed (positive, neutral, or negative).
Product Manager	Person responsible for the development and lifecycle of a product.
Review Text	The full written comment left by a customer.
Rating	Numerical rating (1‑5) provided by a customer.
Overall Sentiment	The summarized sentiment for the whole set of reviews, derived from the average rating.
Sentiment Score	The classification (Positive/Negative) that matches the overall sentiment.
Summary Date	The date on which the review summary was generated.

Appendix C – Reference Materials

1. Sentiment Scale (Rating → Sentiment)

5 – Very Positive
4 – Positive
3 – Neutral
2 – Negative
1 – Very Negative

2. Pros/Cons Extraction Guide

Pros: Look for adjectives or verbs that describe a benefit (e.g., “fast”, “durable”, “easy to use”).
Cons: Look for negative adjectives, complaints, or issues (e.g., “slow”, “broken”, “hard to install”).

3. Summary Formatting Guidelines

Header: Use a short, clear heading (e.g., “Pros Summary”).
Bullet Points: Begin each item with a hyphen (“-”). Keep each point under 20 words.
Consistent Capitalization: Capitalize the first word of each bullet.
No Duplicate Items: If a point appears multiple times, include it only once in the summary.

4. Example Style Guide

Tone: Neutral and direct – focus on facts.
Voice: Use present tense (“The blender is powerful.”).
Grammar: Use complete sentences for the “Key Sentiments” sections.
Length: Keep the entire summary under 200 words for quick reading.

5. Common Errors & How to Address

Issue	Resolution
Missing rating	Flag for manual review; do not include in summary.
Duplicate bullet points	Remove duplicates; keep one entry per unique idea.
Non‑English text	Skip and note “Non‑English review omitted.”
No cons found	Insert “No significant cons identified.”
No pros found	Insert “No notable positives identified.”

6. Sample Data Templates (for user reference)

Customer Review List Template

Product Name: [product name]
Review Date: [YYYY‑MM‑DD]
Reviewer Name: [optional name]
Rating: [1‑5]
Review Text: [full review text]

Review Summary Template

Product Name: [product name]
Summary Date: [YYYY‑MM‑DD]
Overall Sentiment: [Positive/Neutral/Negative]
Average Rating: [X.X]