Review Analytics

Overview

The Reviews section aggregates customer reviews from all connected marketplaces and applies NLP sentiment analysis to surface actionable insights. Instead of reading reviews one by one in each marketplace dashboard, you see patterns across your entire catalog: which products have improving or declining sentiment, which specific topics generate negative feedback, and how your ratings compare to competitors.

The Review Intelligence module extends this with closed-loop listing optimization — automatically generating listing improvements based on review insights.

Key Concepts

Sentiment Score — A score from -1.0 (very negative) to +1.0 (very positive) assigned to each review by the review_sentiment AI worker. The worker uses OpenAI's GPT model to classify sentiment and extract topics.

Topic — A product aspect mentioned in a review. Common topics: battery life, packaging, shipping speed, product quality, value for money, size/fit, customer service. Topics are normalized across languages.

Rating Distribution — The breakdown of 1-star through 5-star reviews. NexusCommerce tracks rating distribution over time to detect deterioration trends.

Review Velocity — The rate of new reviews per week. Sudden increases in review velocity (especially at low ratings) often indicate a product issue.

Getting Started

Navigate to Reviews in the left sidebar. Reviews are imported automatically from connected marketplaces via the adapter sync process.

To trigger a sentiment analysis run on recent reviews:

1. Click Run Sentiment Analysis
2. Select the lookback period (default: last 7 days)
3. Click Start

For continuous automated sentiment analysis, set up a scheduled flow in AI Studio using the review_sentiment job type.

Features

Review Dashboard

The review dashboard shows aggregate metrics across your entire catalog:

• Average rating (all marketplaces, all time vs. selected period)
• Total reviews in selected period
• Sentiment score distribution (positive / neutral / negative)
• Top positive topics (green tags)
• Top negative topics (red tags)
• Products with most new negative reviews (action list)

Per-Product Review View

Click any product from the Products list to see its review data:

Rating Overview:

• Star distribution bar chart
• Average rating trend (90-day line chart)
• Review count trend

Sentiment Timeline:

• 30-day rolling sentiment score
• Marked events (e.g., product changes, shipping carrier changes)

Topic Analysis:

• Extracted topics sorted by frequency
• Per-topic sentiment (positive / neutral / negative)
• Representative review excerpts per topic

Review Feed:

• All reviews sorted by date (newest first)
• Filter by rating, marketplace, sentiment, topic
• Each review shows: rating, date, marketplace, verified purchase status, sentiment score, extracted topics, full review text

Marketplace Comparison

A side-by-side comparison of review data across marketplaces for a specific product:

• Average rating per marketplace
• Review count per marketplace
• Sentiment score per marketplace
• Top topics per marketplace (often differ by region — e.g., German customers comment more on packaging)

Review Alerts

The Reviews section integrates with the Dashboard alert system:

• Rating Drop Alert — Fires when a product's average rating drops more than 0.3 stars in 14 days
• Negative Topic Spike — Fires when a negative topic appears in more than 20% of reviews in a 7-day window
• Review Velocity Spike — Fires when review count in a 24-hour window exceeds 3x the normal daily average

Configure alert thresholds in Settings > Notifications.

Sentiment Analysis Configuration

The review_sentiment worker supports configuration:

• Languages — Specify which review languages to analyze (default: all)
• Topics — Define custom topic categories relevant to your product catalog
• Confidence threshold — Minimum confidence for topic extraction (default: 0.75)

Custom topics example for an electronics seller:

{
  "custom_topics": [
    "battery_life",
    "screen_quality",
    "build_quality",
    "software_stability",
    "charging_speed"
  ]
}

Export

Export review data to CSV for use in your own analysis tools:

GET /api/reviews/export?sku=SKU-001&marketplace=amazon_us&start_date=2026-01-01&format=csv
Authorization: Bearer <token>
X-Tenant-ID: <tenant-id>

Fields exported: review_id, sku, marketplace, rating, date, title, body, sentiment_score, topics, verified_purchase.

Configuration