App Store Complaint Miner

Apify Actor that finds what users hate about competing iOS apps. Point it at 1-5 App Store app IDs and it returns complaint clusters with frequency, severity, missing features, and competitor-switching signals — mined from public customer reviews and clustered by a Groq LLM.

Owner repo: KVKO-Lab/DOCKER-MSaaS · products/app-store-complaint-miner

How it works

1. Resolve each numeric App Store app ID via the iTunes Lookup API → real app name.
2. Collect recent customer reviews via Apple's public RSS feed (itunes.apple.com/{country}/rss/customerreviews/..., up to 500 per app).
3. Cluster the review text with a Groq LLM (qwen/qwen3.8-27b by default) into complaint themes, each tagged with frequency + severity, plus missing features and switching signals — with verbatim quotes as evidence.
4. Emit per-app analysis records, plus a combined report/OUTPUT record.

Input

Field	Type	Default	Notes
apps	array (stringList)	—	1-5 numeric App Store app IDs (required)
country	string	us	Two-letter App Store country code
maxReviewsPerApp	integer	500	50-500 (Apple RSS cap)
minComplaintFrequency	integer	2	Min reviews supporting a complaint cluster
llmApiKey	string (secret)	env GROQ_API_KEY	Groq key from console.groq.com
llmModel	string	qwen/qwen3.8-27b	Groq chat model

Output

Three kinds of records pushed to the dataset:

• { kind: "app-reviews", ... } — raw fetched reviews per app.
• { kind: "app-analysis", ... } — per-app summary, complaints[] (title/frequency/severity/quotes), missingFeatures[], switchingSignals[].
• { kind: "report", ... } — full aggregate; also stored in the default Key-Value store as OUTPUT.

Development

npm install
npm run type-check   # tsc --noEmit
npm run build        # tsc -> dist
npm start            # node dist/main.js (needs a default local input)

Local run needs an Apify input — create .actor/INPUT.json (see Apify CLI docs) or set APIFY_LOCAL_STORAGE_DIR and the standard storage layout.

Stack

TypeScript, apify SDK, OpenAI-compatible Groq client. No browser needed. Base image: apify/actor-node:22.