DEV Community

Ramazan Akdoğan
Ramazan Akdoğan

Posted on Edited on

Your users are telling you why they churn. Here's how to read 6,000 App Store reviews in 8 seconds

Your app shipped an update on Tuesday. On Thursday, the 1-star reviews start coming in.

"Crashes every time I open it." "Why is this a subscription now??" "The new layout is unusable."

The reviews are there, but they're spread over 175 country storefronts, Apple shows only 500 per country, and most tools make you look up a numeric ID and scrape one country at a time. So nobody reads them, and the churn continues.

I got tired of that and built a tool that collects thousands of reviews from 120+ countries in one run. It took 8 seconds to get about 6,000 Spotify reviews.

Meet the App Store Reviews Scraper

👉 App Store Reviews Scraper

  • 🔎 Just type the app name. "Duolingo" works, and so do URLs and IDs. No ID lookup needed.
  • 🌍 One run, the whole world. Set countries: ["all"] and every storefront is scanned in parallel.
  • 🎯 Laser filters. Only 1–2★, only since your last release, only reviews that mention crash, subscription or ads.
  • 🧠 Analysis-ready rows. Each review comes with the app name, developer, genre, current version, average rating, and the version the user reviewed, so you can tie complaints to a specific release.
  • ⚡ Reliable. It's built on Apple's public review feed and the official iTunes Search API. There's no login, no proxies and no headless browser that breaks next month.

From raw reviews to a roadmap in 15 lines

from apify_client import ApifyClient
import pandas as pd

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("datacraft-tools/app-store-reviews").call(run_input={
    "apps": ["Duolingo"],
    "countries": ["us", "gb", "ca", "au", "de"],
    "maxReviewsPerCountry": 500,
    "maxRating": 2,                 # complaints only
    "reviewsSince": "2026-11-01",   # since the last release
})
df = pd.DataFrame(client.dataset(run["defaultDatasetId"]).list_items().items)

print(df.groupby("reviewedVersion").size().sort_values(ascending=False).head())   # which release hurts?
print(df["text"].str.lower().str.extract(r"(crash|freeze|ads|subscription|price|bug)")[0].value_counts())
Enter fullscreen mode Exit fullscreen mode

Now paste df["text"] into ChatGPT or Claude with:

"Group these complaints into themes, rank them by frequency and suggest one fix for each."

You get a prioritized backlog written by your own users, before your weekly standup.

Three more ways people use it

🥊 Competitor teardown. Run it on your 5 biggest competitors and compare their 1★ themes. Their users' top complaint is your next landing-page headline.

🌐 Localization radar. If the app is rated 4.6★ in the US and 3.1★ in Germany, there's probably a broken translation, a pricing problem or a payment bug. You'll see it within minutes.

📈 ASO gold. Users describe your app in their own words. Mine the most frequent phrases and put them in your keywords and screenshots.

Pricing

$0.20 per 1,000 reviews plus $0.01 per run. A full 5-country teardown of a competitor costs about 50 cents, less than a coffee and far cheaper than another month of churn.

👉 Try it on your app now. Your first run takes about a minute to set up.


I built this actor and I read every comment. Tell me which feature you'd want next: Google Play? automatic sentiment? weekly email alerts?

Top comments (0)