App Review Insights

Type a product name and this Actor finds it across the Apple App Store, Google Play, Capterra, and Trustpilot, pulls the reviews into one unified dataset, and turns them into an actionable, executive-ready report — prioritized complaints, key strengths, feature requests, sentiment and rating trends.

No review IDs, no per-source configuration: it resolves the product by name on every source automatically. Each source is fail-soft — if one is unavailable, the others still produce a full report.

Two modes

	Free (default)	AI analysis (optional)
Cost	Free	Your own OpenAI API key
Engine	Local rule-based (VADER sentiment + aspect taxonomy)	OpenAI, grounded on the same reviews + stats
Output	Ratings, sentiment, aspect-based complaints/praises ranked by severity, feature requests, monthly trend, representative quotes	Everything in Free plus an AI analysis section: plain-language verdict, prioritized issues with recommendations, strengths, opportunities and quotes

• Free mode runs whenever no key is supplied. It groups reviews into named aspects (e.g. Meetings & Calls, Login & Accounts, Performance & Reliability), ranks complaints by how many unhappy customers they affect (volume x negativity), and picks a representative, sentiment-matched quote for each.
• AI mode activates only when an OpenAI API key is provided. It samples a balanced set of reviews, sends them with the pre-computed statistics to OpenAI, and returns a concise product-team briefing. It is fail-soft: if the API call fails, the free report is still produced. Your key is used only for the run and is never written to the output.

Sources

Source	How it is fetched
Apple App Store	iTunes Search API (resolution) + server-rendered web reviews page
Google Play	google-play-scraper
Capterra	HTTP + Beautiful Soup
Trustpilot	Headless Chromium via patchright

Input

Field	Key	Required	Default	Description
Product name	productName	Yes	—	The product to analyze (e.g. Notion, Microsoft Teams).
App Store country	appStoreCountry	No	us	Two-letter storefront code for App Store search/reviews.
Max reviews per source	maxReviewsPerSource	No	200	Upper bound on reviews fetched per source (10–500).
OpenAI API key	openaiApiKey	No	—	Optional. Enables the AI analysis section. Marked secret.
OpenAI model	openaiModel	No	gpt-4o-mini	Model for the AI analysis (only used when a key is given).

Example input:

{
  "productName": "Microsoft Teams",
  "appStoreCountry": "us",
  "maxReviewsPerSource": 100,
  "openaiApiKey": "sk-...",
  "openaiModel": "gpt-4o-mini"
}

Output

• Dataset — every fetched review in a unified shape (source, product, rating, title, body, author, date, url, ...).
• Key-value store
  • OUTPUT — the full insights object as JSON (includes aiAnalysis when AI mode ran).
  • REPORT.md — a shareable Markdown report (the AI section appears first when enabled).

Run locally

Requires Python 3.11+ (the deployed image uses 3.13).

python -m venv .venv
.venv\Scripts\activate            # Windows
# source .venv/bin/activate       # macOS / Linux
pip install -r requirements.txt
python -m patchright install chromium   # for Trustpilot

Put your input in storage/key_value_stores/default/INPUT.json, then run:

$python -m my_actor

Or, using the Apify CLI (reads the same INPUT.json):

$apify run

The report is written to storage/key_value_stores/default/REPORT.md.

To try AI mode locally, either add "openaiApiKey": "sk-..." to INPUT.json, or set the environment variable OPENAI_API_KEY before running.

Deploy to Apify

Prerequisites

• An Apify account (free tier is fine).

• The Apify CLI:

  npm install -g apify-cli
  # or: brew install apify-cli
  

• Log in with your API token:

  $apify login
  

Option A — Push from your machine (fastest)

From the project root (app-review-insights/, the folder containing .actor/):

$apify push

This uploads the source, builds the Docker image on Apify (installs requirements.txt and Chromium per the Dockerfile), and creates the Actor under Actors → My Actors. Re-run apify push to deploy updates.

The first build takes a few minutes because it installs the Playwright/Chromium base image and dependencies. Watch the build log in the console; a green Succeeded means it is ready.

Option B — Connect a Git repository (best for ongoing work)

1. Push this project to GitHub/GitLab/Bitbucket.
2. Go to the Actor creation page and click Link Git Repository.
3. Point it at your repo (and the folder with .actor/ if it is a monorepo).
4. Apify rebuilds automatically on every push to the tracked branch.

Run it on the platform

1. Open the Actor and go to the Input tab — the fields above render automatically from the input schema.
2. Enter a Product name. To enable AI mode, paste your OpenAI API key (it renders as a secret field).
3. Click Start.
4. When it finishes:
   • Storage → Dataset holds the raw reviews (exportable as JSON/CSV/Excel).
   • Storage → Key-value store holds OUTPUT (JSON) and REPORT.md (the readable report).

Notes on cost and secrets

• The Actor itself needs no paid API keys. Only AI mode uses your OpenAI key, and OpenAI bills you directly for those tokens.
• On the platform the OpenAI key is stored as a secret input and is not saved into the Actor output.
• You can schedule runs, call the Actor via the API, or wire it into integrations (Make, Zapier, Slack, Google Drive, ...).

Project structure

my_actor/
  main.py            # orchestration: resolve, fetch, analyze, report
  insights.py        # free rule-based analysis (VADER + aspect taxonomy)
  ai.py              # optional OpenAI analysis (only runs with a key)
  report.py          # Markdown report renderer
  sources/           # app_store, play_store, capterra, trustpilot
.actor/              # actor.json, input/output/dataset schemas
Dockerfile           # Python + Playwright base image
requirements.txt

Documentation reference

• Apify SDK for Python
• Apify Platform documentation
• Actor input schema
• Deploying Actors