Google Maps Review Intelligence Analyzer

Turn customer reviews into business intelligence.

This Actor is not another review scraper. It's the AI analysis layer that sits on top of any scraper. Feed it reviews from Google Maps, Yelp, TripAdvisor, Booking.com, or any source, and get back:

• Per-review sentiment scores (-1.0 to +1.0) and classifications
• Topic-level sentiment (what customers love and hate, and by how much)
• Key phrases extracted verbatim from each review
• One actionable business insight per review
• A full reputation report with strengths, weaknesses, trend analysis, and concrete recommendations

---

How It Works

Step 1: Run any review scraper on the Apify Store (or bring your own data).

Step 2: Pass the dataset ID (or raw JSON) to this Actor.

Step 3: Get AI-enriched reviews + a reputation intelligence report.

The Actor auto-detects field names from any scraper output. No manual mapping needed.

---

Example Output

Per-Review (dataset)

{
    "placeName": "Joe's Pizza",
    "reviewerName": "Sarah M.",
    "stars": 4,
    "text": "Great pizza but the wait was too long. Staff were friendly though.",
    "sentiment": "mixed",
    "sentimentScore": 0.42,
    "topics": [
        { "topic": "food quality", "sentiment": "positive", "score": 0.9 },
        { "topic": "wait time", "sentiment": "negative", "score": -0.7 },
        { "topic": "staff friendliness", "sentiment": "positive", "score": 0.8 }
    ],
    "keyPhrases": ["great pizza", "wait was too long", "staff were friendly"],
    "actionableInsight": "Food quality praised but wait times are a recurring pain point"
}

Reputation Report (key-value store: REPORT)

{
    "placeName": "Joe's Pizza",
    "totalReviewsAnalyzed": 200,
    "overallSentiment": "positive",
    "overallSentimentScore": 0.65,
    "averageStars": 4.2,
    "sentimentDistribution": {
        "positive": 58.5,
        "negative": 18.0,
        "neutral": 12.5,
        "mixed": 11.0
    },
    "topicScores": {
        "food quality": { "score": 0.82, "mentionCount": 145, "mentionRate": 72.5, "sentiment": "positive" },
        "wait time": { "score": -0.55, "mentionCount": 89, "mentionRate": 44.5, "sentiment": "negative" }
    },
    "strengths": [
        { "topic": "food quality", "score": 0.82, "mentions": 145, "summary": "..." }
    ],
    "weaknesses": [
        { "topic": "wait time", "score": -0.55, "mentions": 89, "summary": "..." }
    ],
    "recommendations": [
        "Reduce wait times with queue management or reservations",
        "Only 30% of negative reviews have owner responses. Responding shows customers you care.",
        "Your strongest area is food quality. Feature this in marketing materials."
    ],
    "responseAnalysis": {
        "totalResponseRate": 45.0,
        "negativeResponseRate": 30.0
    },
    "reviewVelocity": { "trend": "increasing", "changePercent": 15.2 },
    "topPhrases": [
        { "phrase": "great pizza", "count": 23 },
        { "phrase": "long wait", "count": 18 }
    ],
    "sentimentTrend": [
        { "month": "2026-01", "sentimentScore": 0.62, "averageStars": 4.1, "reviewCount": 45 }
    ]
}

---

Two Input Modes

Mode 1: Apify Dataset ID (recommended)

Run any review scraper first, then pass its dataset ID. No re-scraping needed.

{
    "reviewDatasetId": "abc123xyz"
}

Compatible scrapers (any output with text + rating fields works):

• Google Maps Reviews Scraper
• Yelp Scraper
• TripAdvisor Scraper
• Booking.com Reviews
• Any scraper that outputs text/reviewText and stars/rating fields

Mode 2: Raw JSON (for testing or one-off analysis)

{
    "reviewData": [
        { "text": "Amazing food, terrible service.", "stars": 2 },
        { "text": "Hidden gem. Best ramen I've had.", "stars": 5 }
    ]
}

---

Business Type Presets

Select your business type to get optimized topic categories:

Preset	Topics
Restaurant	food quality, service, price/value, cleanliness, atmosphere, location, parking, wait time, portions, staff
Hotel	room cleanliness, check-in, bed comfort, breakfast, wifi, noise, location, pool/amenities, staff, value
Salon	skill/quality, wait, cleanliness, staff, pricing, atmosphere, location, booking, products, results
Clinic	doctor competence, wait time, staff, cleanliness, booking, billing, diagnosis, follow-up, facility, communication
Retail	product quality, variety, pricing, staff, cleanliness, layout, checkout, returns, location, parking
Custom	Define your own 5-15 topics

---

What the Report Tells You

Section	Business Question It Answers
sentimentDistribution	What % of my reviews are positive vs negative?
topicScores	Which aspects of my business do customers care about most?
strengths	What should I double down on in marketing?
weaknesses	What's hurting my rating the most?
recommendations	What specific actions should I take?
responseAnalysis	Am I responding to negative reviews enough?
reviewVelocity	Are my reviews increasing or declining?
sentimentTrend	Is customer satisfaction improving or getting worse?
topPhrases	What words do customers use most? (word cloud data)
topPositiveQuotes / topNegativeQuotes	What are my best and worst reviews?

When analyzing multiple places, the Actor also generates a COMPARISON report ranking all places by sentiment score with their top strengths and weaknesses side by side.

---

AI Providers

Provider	Default Model	Setup
Built-in AI	GPT-5.4-mini	No key needed
OpenAI	GPT-5.4-mini	Your key from platform.openai.com
Anthropic	Claude Sonnet 4	Your key from console.anthropic.com
OpenRouter	GPT-4o-mini	Your key from openrouter.ai/keys

Built-in AI has zero setup. For high-volume runs (1000+ reviews), bring your own key to save on costs.

---

Recommended Workflow

1. Run a review scraper (e.g. Google Maps Reviews Scraper)
2. Copy the dataset ID from the completed run
3. Paste it into this Actor's reviewDatasetId field
4. Select your business type and click Start
5. Check the Dataset tab for enriched reviews (sentiment, topics, insights per review)
6. Check the Key-value store tab for the full reputation report (see below)

For recurring monitoring, schedule both the scraper and this Actor to run monthly.

---

How to Access the Reputation Report

The aggregated report is stored in the Key-value store, not in the Dataset. The Dataset contains per-review analysis. The report is the big-picture summary.

To find it after a run completes:

1. Go to your completed run in Apify Console
2. Click the Key-value store tab (next to Dataset, Log, Info)
3. Click the REPORT key to view the full JSON report
4. Click Download to save it as a JSON file

If you analyzed multiple places in one run, each place gets its own report key (e.g. REPORT_JOES_PIZZA, REPORT_MAMA_MIA) plus a COMPARISON key ranking all places side by side.

Tip: You can also access the report programmatically via the Apify API:

GET https://api.apify.com/v2/key-value-stores/{storeId}/records/REPORT

---

Field Auto-Detection

The Actor automatically recognizes review fields from any scraper output:

Our Field	Detected From
Review text	text, reviewText, review, snippet, body, comment, content
Star rating	stars, rating, score, reviewRating, totalScore (auto-scales 1-10 to 1-5)
Date	publishedAt, publishedAtDate, date, iso_date, createdAt
Reviewer	reviewerName, authorName, author, user.name
Place name	placeName, title, name, businessName
Owner response	ownerResponse, responseFromOwnerText, response.snippet

No manual field mapping needed. Just pipe in any dataset.

---

Environment Variables (Actor Owners)

Variable	Value	Notes
BUILT_IN_API_KEY	Your OpenAI API key (sk-...)	Set as secret in Apify Console

---

Limitations

• Reviews under 10 characters are skipped for AI analysis (still counted in rating distributions)
• Review text is capped at 2000 characters to control AI costs
• AI analysis quality depends on review text quality
• Sentiment scores are AI-generated approximations, not ground truth

---

Changelog

v1.0

• Two input modes: Apify dataset ID, raw JSON
• Auto-detection of review fields from any scraper output
• Multi-provider AI: Built-in, OpenAI, Anthropic, OpenRouter
• Business type presets: restaurant, hotel, salon, clinic, retail, custom
• Per-review: sentiment, topic scoring, key phrases, actionable insights
• Reputation report: sentiment distribution, topic analysis, owner response rate, review velocity, monthly trends, phrase frequency, recommendations
• Multi-place comparison reports
• Retry with exponential backoff on AI API failures