Multi-Platform Review Aggregator

Aggregate reviews from Google, Yelp, TripAdvisor, and Trustpilot into a unified dataset with AI-powered sentiment analysis and authenticity scoring.

Features

• Multi-platform aggregation: Collect reviews from 4 major platforms in a single run
• Unified schema: Consistent output format across all platforms
• AI sentiment analysis: GPT-powered sentiment classification with aspect-based breakdown
• Authenticity scoring: Detect potential fake reviews using heuristic signals
• Flexible input: Search by business name or provide specific URLs/IDs

Use Cases

• Reputation monitoring: Track what customers say across platforms
• Competitive analysis: Compare your reviews against competitors
• Market research: Understand customer sentiment in a market segment
• Lead qualification: Assess business quality before partnership
• Content analysis: Extract themes and pain points from reviews

Platforms Supported

Platform	Data Source	Features	Reliability
Google	Google Maps Reviews	Reviews, owner responses, images	⭐⭐⭐⭐⭐ High
Yelp	Yelp Business Pages	Reviews, ratings, business info	⭐⭐ Low*
TripAdvisor	TripAdvisor	Reviews, trip type, travel date	⭐⭐⭐⭐ Good
Trustpilot	Trustpilot.com	Reviews, verification status	⭐⭐⭐⭐⭐ High

*Yelp uses aggressive anti-bot measures. CSS selectors may break without notice. Residential proxies strongly recommended. Expect partial or failed extraction in many cases.

Example Input

Search by business name

{
  "businessName": "Shake Shack New York",
  "platforms": ["google", "yelp", "tripadvisor"],
  "maxReviewsPerPlatform": 50,
  "enableSentimentAnalysis": true
}

Use specific URLs

{
  "googleMapsUrl": "https://maps.google.com/maps?cid=1234567890",
  "yelpUrl": "https://www.yelp.com/biz/shake-shack-new-york",
  "tripadvisorUrl": "https://www.tripadvisor.com/Restaurant_Review-...",
  "trustpilotDomain": "shakeshack.com",
  "platforms": ["google", "yelp", "tripadvisor", "trustpilot"],
  "maxReviewsPerPlatform": 100,
  "enableSentimentAnalysis": true,
  "openaiApiKey": "sk-..."
}

Input Parameters

Business Identification

Parameter	Type	Description
businessName	string	Business name to search across platforms (used when specific URLs not provided)
googlePlaceId	string	Google Maps Place ID for precise matching
googleMapsUrl	string	Direct Google Maps URL to the business
yelpUrl	string	Direct Yelp business page URL
tripadvisorUrl	string	Direct TripAdvisor page URL
tripadvisorLocationId	string	TripAdvisor location ID (alternative to URL)
trustpilotDomain	string	Business domain on Trustpilot (default: amazon.com)
trustpilotUrl	string	Direct Trustpilot URL (e.g., trustpilot.com/review/example.com)

Scraping Options

Parameter	Type	Default	Description
platforms	array	["trustpilot"]	Platforms to scrape: google, yelp, tripadvisor, trustpilot
maxReviewsPerPlatform	number	20	Max reviews per platform (1-500)
language	string	"en"	Preferred language (ISO 639-1 code)

AI Features

Parameter	Type	Default	Description
enableSentimentAnalysis	boolean	true	Analyze sentiment (uses OpenAI if key provided, falls back to rule-based)
enableAuthenticityScoring	boolean	true	Detect potential fake reviews using heuristic signals
openaiApiKey	string	-	OpenAI API key for AI-powered sentiment (~$0.01/100 reviews)

Advanced

Parameter	Type	Description
proxyConfiguration	object	Proxy settings for scraping protected sites (recommended for Yelp)

Output Schema

Review Record

Each review includes:

Field	Description
platform	Source platform (google, yelp, tripadvisor, trustpilot)
entityName	Business name
entityType	Business type (restaurant, hotel, local_business, etc.)
text	Full review text
title	Review title (if available)
starRating	Rating (1-5 stars)
normalizedRating	Rating normalized to 0-1
reviewerName	Reviewer's display name
datePublished	Review date (ISO 8601)
ownerResponse	Business owner's reply (if any)
images	Array of attached image URLs (Google, TripAdvisor only; Yelp/Trustpilot return empty array)
sentiment	AI sentiment analysis (see below)
authenticitySignals	Fake review detection signals

Sentiment Object

With OpenAI API key (AI-powered):

{
  "overall": "positive",
  "score": 0.85,
  "confidence": 0.92,
  "aspects": [
    {"aspect": "service", "sentiment": "positive", "score": 0.9, "snippet": "staff was amazing"},
    {"aspect": "price", "sentiment": "negative", "score": -0.3, "snippet": "bit pricey"}
  ]
}

Without API key (rule-based fallback):

{
  "overall": "positive",
  "score": 0.6,
  "confidence": 0.6,
  "aspects": [],
  "method": "rule-based"
}

Note: Rule-based analysis has lower confidence (0.6) and does not provide aspect breakdown.

Authenticity Signals

{
  "score": 0.85,
  "flags": ["short_text", "extreme_rating_no_details"],
  "flagCount": 2
}

Flags include:

• very_short_text - Review under 20 characters
• short_text - Review under 50 characters
• extreme_rating_no_details - 1 or 5 star without details
• generic_language - Overuse of common phrases
• excessive_caps - Too many capital letters
• excessive_exclamations - More than 5 exclamation marks
• rating_sentiment_mismatch - Text sentiment doesn't match rating
• single_review_account - Reviewer has only one review
• no_specific_details - Review lacks specific details (dates, numbers, names)

Summary Record

Each run also outputs an aggregate summary:

{
  "dataType": "summary",
  "entityName": "Shake Shack",
  "totalReviews": 150,
  "averageRating": 4.2,
  "ratingDistribution": {"1": 5, "2": 8, "3": 15, "4": 52, "5": 70},
  "sentimentBreakdown": {
    "positive": 0.65,
    "negative": 0.15,
    "neutral": 0.12,
    "mixed": 0.08
  },
  "topPositiveAspects": [
    {"aspect": "food_quality", "avgScore": 0.82, "frequency": 89}
  ],
  "topNegativeAspects": [
    {"aspect": "wait_time", "avgScore": -0.45, "frequency": 34}
  ],
  "platformBreakdown": {"google": 50, "yelp": 50, "tripadvisor": 50}
}

AI Sentiment Analysis

With OpenAI API Key

• Aspect-based sentiment using GPT-4o-mini
• Identifies specific topics (food, service, price, etc.)
• High accuracy sentiment classification
• ~$0.01 per 100 reviews

Without API Key (Rule-Based Fallback)

• Keyword-based sentiment scoring
• Combined with rating signal
• Lower accuracy but free
• No aspect breakdown
• Output includes "method": "rule-based" field

Pricing

This actor uses Apify compute units plus optional external dependencies:

Component	Cost
Google Reviews	Uses compass/google-maps-reviews-scraper (paid actor)
TripAdvisor	Uses maxcopell/tripadvisor (paid actor)
Yelp, Trustpilot	Direct scraping (compute units only)
AI Sentiment	OpenAI API (~$0.01/100 reviews)

Key-Value Store Schema

The actor stores the following data in the key-value store:

RUN_STATS

{
  "google": 50,
  "yelp": 45,
  "tripadvisor": 48,
  "trustpilot": 50,
  "total": 193,
  "sentimentAnalyzed": true,
  "authenticityScored": true
}

Tips

1. Start with Google + Trustpilot - These are the most reliable platforms
2. Use specific URLs when possible for accurate matching
3. Yelp scraping is unreliable - Expect failures; use residential proxies and consider Yelp optional
4. AI sentiment is optional - Rule-based fallback works for basic needs
5. Validate OpenAI key first - Invalid keys fail silently mid-run; test before large jobs
6. Language filter - Works for Google/TripAdvisor/Trustpilot; Yelp ignores this parameter

Support

For issues or feature requests, please contact the author through the Apify platform.

---

Built by JCD | Powered by Apify