How to analyze app reviews: a simple process for App Store and Google Play

Your App Store and Google Play reviews are the cheapest user research you will ever get. People tell you, in their own words, what broke, what confused them and what they would pay for. The hard part isn't getting feedback. It's turning hundreds of short, angry, multilingual messages into a list of things to fix.

This guide shows a simple process indie developers can run in an hour a week, by hand or with a tool.

1. Collect every review in one place

App Store Connect and Google Play Console each show reviews for one store, one app at a time, and neither makes it easy to export them. Before you analyze anything, get all reviews into one list: a spreadsheet works for a few dozen a month.

For every review keep at least:

• rating, date and text
• store and country
• app version (it tells you which release caused a complaint)
• whether you replied

Both stores have official APIs for this. See how to create an App Store Connect API key and a Google Play service account. Note that the Google Play API only returns reviews from the last 7 days, so start saving them early.

2. Tag sentiment and topic

The star rating alone is misleading. A 4★ review can hide a serious bug, and a 1★ review can be about a payment problem that has nothing to do with your app. Give every review two labels:

• Sentiment: positive, neutral, negative or mixed. “Mixed” matters most: “Love it, but it crashes when I export” is a happy user about to leave.
• Topic: bug, feature request, UX, performance, pricing, content or praise. Keep the list short and the same every week, so you can compare.

3. Group reviews into issues

This is the step that turns feedback into a roadmap. Read the negative and mixed reviews and give each one a short issue name that describes the problem, not the user. Reuse the same name when another review says the same thing.

• Good: “Widget doesn't update after midnight”, “Crash when opening the camera”, “Can't restore purchase on new phone”.
• Bad: “Bug”, “User unhappy”, “Bad review”.

After a week, count reviews per issue. Twelve people writing about the same crash, in four languages, is a clear priority. One person asking for a dark mode is an idea for later.

4. Compare by version and week

Put the numbers next to your releases. If “Timer resets in background” appears only in reviews for v3.1, you know which change to look at. Track the weekly average rating too: a drop of half a star in a week is a signal worth acting on today. Here's what to do when your rating drops.

5. Act and close the loop

1. Fix the top issue

   Pick the issue with the most recent reviews, not the loudest single review.

2. Reply to everyone who reported it

   Tell them it's fixed and in which version. Reviewers get a notification and often update their rating. Our 30 reply templates help here.

3. Watch the issue count fall

   If new reviews still mention it after the fix, the fix didn't work.

When there are too many reviews to read

Around 50 reviews a week, manual tagging stops being fun. Around 200, it stops happening. That's the point where AI review analysis pays off: a model can tag sentiment and topic, translate, write a one-line summary and assign each review to an issue, and you only read the short list. Keep a human on the last step: decide what to fix and what to say yourself.

Bonus: read your competitors' reviews

Competitor reviews are public on their store pages, and they are a free list of what users want and can't get. Read the 1★ to 3★ reviews of the top three apps in your category and tag them with the same topics. Repeated complaints you don't share are your positioning; repeated requests nobody solves are your next feature.