🔬 App Review Pain Miner - AI Product Intelligence from User Reviews

Turn thousands of app reviews into actionable product intelligence in minutes. This actor ingests user reviews from any source -- inline JSON, CSV uploads, or live web scraping -- then runs NLP analysis to cluster complaints, score opportunities, and generate a prioritized product roadmap and sales outreach brief. All without any API keys required. Zero config, zero cost surprises.

Whether you're a product manager mining your own app's reviews, a SaaS founder researching competitors, or an agency building pitches for prospective clients, Pain Miner gives you the structured intelligence that takes analysts days to produce manually.

---

📊 What Data Can You Extract?

• Pain point clusters -- Similar complaints grouped together with frequency counts, severity scores, and representative quotes
• Opportunity scores -- Each cluster scored across 7 factors: frequency, severity, recency, monetization potential, churn risk, competitive gap, and reply gap
• Product roadmap -- Prioritized list of fixes/features ranked by opportunity score with effort estimates
• Sales outreach brief -- Pre-written talking points, subject lines, and pain-point references for cold outreach to the app's team
• Expert debate -- Simulated product strategy debate analyzing trade-offs of different approaches
• Summary statistics -- Review volume, sentiment distribution, trend analysis, top keywords

---

⚡ Key Features

Feature	Description
Zero Config	Works without any API keys -- pure heuristic NLP analysis out of the box
Multi-Source Input	Inline JSON array, CSV file URL, or live web scraping from app stores and review sites
Smart Clustering	Groups similar complaints using token-set similarity, not just keyword matching
7-Factor Opportunity Scoring	Frequency, severity, recency, monetization potential, churn risk, competitive gap, reply gap
4 Output Documents	Summary, roadmap, outreach brief, expert debate -- all structured JSON
Optional LLM Enhancement	Bring your own key for OpenAI, Gemini, Groq, or OpenRouter to get richer natural-language insights
Three Analysis Modes	fast (heuristics only), balanced (heuristics + light LLM), deep (full LLM-enhanced analysis)

---

📥 Input

Parameter	Type	Required	Default	Description
reviews	Array	No*	--	Inline array of review objects {text, rating, date, author}
reviewUrls	Array	No*	--	URLs to scrape reviews from (app store pages, G2, Capterra, etc.)
reviewCsvUrl	String	No*	--	URL to a CSV file containing reviews
analysisMode	String	No	balanced	fast (heuristics only), balanced, or deep (requires LLM key)
provider	String	No	--	LLM provider: openai, gemini, groq, openrouter
apiKey	String	No	--	Your LLM API key (BYOK -- we never store it)
model	String	No	--	Specific model to use (e.g., gpt-4o, gemini-2.0-flash)
maxReviews	Integer	No	1000	Maximum reviews to analyze

*At least one of reviews, reviewUrls, or reviewCsvUrl is required.

Example input (zero-config mode):

{
  "reviewUrls": [
    "https://apps.apple.com/us/app/slack/id618783545?see-all=reviews",
    "https://play.google.com/store/apps/details?id=com.Slack&showAllReviews=true"
  ],
  "analysisMode": "fast",
  "maxReviews": 500
}

Example input (LLM-enhanced mode):

{
  "reviews": [
    {"text": "App crashes every time I try to upload a photo", "rating": 1, "date": "2026-02-15"},
    {"text": "Photo upload is broken again after the last update", "rating": 1, "date": "2026-02-20"},
    {"text": "Love the new UI but notifications are way too aggressive", "rating": 3, "date": "2026-02-18"}
  ],
  "analysisMode": "deep",
  "provider": "openai",
  "apiKey": "sk-...",
  "model": "gpt-4o"
}

---

📤 Output Example

The actor produces 4 output files in the key-value store:

summary.json

{
  "appName": "Slack",
  "totalReviewsAnalyzed": 487,
  "averageRating": 2.8,
  "sentimentBreakdown": {
    "positive": 124,
    "neutral": 89,
    "negative": 274
  },
  "topPainClusters": [
    {
      "id": "cluster_01",
      "label": "Notification overload / inability to mute channels",
      "frequency": 67,
      "severityScore": 8.4,
      "opportunityScore": 91.2,
      "representativeQuotes": [
        "I get hundreds of notifications a day and there's no way to batch-mute channels",
        "The Do Not Disturb feature resets every morning, completely useless",
        "I had to uninstall because notifications were killing my productivity"
      ],
      "firstSeen": "2025-09-12",
      "lastSeen": "2026-02-28",
      "trend": "worsening"
    },
    {
      "id": "cluster_02",
      "label": "Search is slow and misses results",
      "frequency": 52,
      "severityScore": 7.1,
      "opportunityScore": 78.5,
      "representativeQuotes": [
        "Search takes 10+ seconds and half the time doesn't find messages I know exist",
        "Tried searching for a file someone shared last week, nothing came up"
      ],
      "firstSeen": "2025-11-03",
      "lastSeen": "2026-02-27",
      "trend": "stable"
    }
  ],
  "trendingKeywords": ["notifications", "search", "battery", "crash", "slow"],
  "analysisMode": "fast",
  "processingTime": "34s"
}

roadmap.json

{
  "prioritizedItems": [
    {
      "rank": 1,
      "clusterId": "cluster_01",
      "title": "Overhaul notification controls",
      "description": "Add batch-mute, persistent DND schedules, and per-channel notification granularity",
      "opportunityScore": 91.2,
      "estimatedEffort": "medium",
      "impactedUsers": "~14% of reviewers",
      "suggestedApproach": "Ship granular per-channel mute as quick win, then build scheduled DND as fast-follow"
    },
    {
      "rank": 2,
      "clusterId": "cluster_02",
      "title": "Improve search speed and relevance",
      "description": "Address slow search performance and missing results in message/file search",
      "opportunityScore": 78.5,
      "estimatedEffort": "high",
      "impactedUsers": "~11% of reviewers"
    }
  ]
}

outreach_brief.json

{
  "targetCompany": "Slack Technologies",
  "generatedAt": "2026-03-02T10:15:00Z",
  "talkingPoints": [
    {
      "painPoint": "Notification overload",
      "frequency": 67,
      "sampleQuote": "I had to uninstall because notifications were killing my productivity",
      "suggestedAngle": "Users are churning specifically because notification controls are too coarse-grained. 14% of negative reviews cite this."
    }
  ],
  "subjectLines": [
    "67 Slack users this month complained about the same thing",
    "The #1 reason Slack gets 1-star reviews (it's not bugs)"
  ],
  "briefSummary": "Notification management is Slack's most-cited pain point in recent reviews, with complaints worsening over the last 6 months. Search reliability is the second biggest cluster. Both represent high-impact opportunities."
}

---

💰 Pricing

This actor uses Pay Per Event (PPE) pricing:

Event	Cost
Per review analyzed	$0.01

Volume	Cost
100 reviews	$1.00
500 reviews	$5.00
1,000 reviews	$10.00
5,000 reviews	$50.00

Note: LLM costs (if you enable balanced or deep mode with your own API key) are billed directly by your LLM provider, not by this actor. In fast mode, there are zero external API costs.

---

🎯 Use Cases

• Product managers -- Mine your own app's reviews to find the most impactful bugs and feature requests, prioritized by real user pain
• Competitive intelligence -- Analyze a competitor's reviews to find their weaknesses before they do, and position your product accordingly
• Agency pitches -- Generate a data-backed outreach brief showing a prospect exactly what their users hate, complete with quotes and frequency counts
• Investor due diligence -- Quickly assess product-market fit and user satisfaction trends for portfolio companies or acquisition targets

---

🔌 API Usage

curl "https://api.apify.com/v2/acts/YOUR_USERNAME~app-review-pain-miner/runs" \
  -X POST \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -d '{
    "reviewUrls": [
      "https://play.google.com/store/apps/details?id=com.example.app&showAllReviews=true"
    ],
    "analysisMode": "fast",
    "maxReviews": 500
  }'

Retrieve the summary from the key-value store:

# Get the default key-value store ID from the run
RUN_ID="YOUR_RUN_ID"
STORE_ID=$(curl -s "https://api.apify.com/v2/actor-runs/$RUN_ID" \
  -H "Authorization: Bearer YOUR_API_TOKEN" | jq -r '.data.defaultKeyValueStoreId')

# Download summary
curl "https://api.apify.com/v2/key-value-stores/$STORE_ID/records/summary.json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -o summary.json

# Download roadmap
curl "https://api.apify.com/v2/key-value-stores/$STORE_ID/records/roadmap.json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -o roadmap.json

---

❓ FAQ

Q: Do I need an OpenAI/Gemini API key to use this? A: No. The fast analysis mode uses pure heuristic NLP -- no external API calls, no API keys, no additional costs. The LLM-enhanced modes (balanced and deep) are optional upgrades that use your own API key for richer natural-language summaries and more nuanced clustering.

Q: What review sources can it scrape? A: The actor can scrape reviews from Apple App Store, Google Play Store, and most public review pages. You can also provide reviews directly as a JSON array or CSV file URL, which means you can feed it data from any source -- G2, Capterra, Trustpilot, Amazon, internal feedback databases, support tickets, or survey responses.

Q: How does the opportunity scoring work? A: Each pain cluster is scored across 7 weighted factors: frequency (how often it's mentioned), severity (how angry the reviewers are), recency (is it getting worse?), monetization potential (does it affect paid users?), churn risk (are people threatening to leave?), competitive gap (do competitors solve this?), and reply gap (has the developer responded?). The composite score ranges from 0-100 and directly feeds the prioritized roadmap.

---

🛟 Support

• Bug reports & feature requests: Open an issue on the actor's GitHub repository or use the Apify Console issue tracker
• Questions: Post in the actor's discussion tab on Apify Store
• Custom analysis pipelines: Contact us for help integrating Pain Miner into your product analytics workflow