1. Your reviews export
Needs a rating column and a review text column. A date, an app version and a developer reply column unlock the month, release and reply views.
Choose a file or paste CSV text, then click Analyze.
Result
Complaint themes, by stars they cost
| Theme | Reviews | Share | Their average | Overall average if fixed |
|---|
By version
| Version | Reviews | Average | Unhappy share |
|---|
By month
| Month | Reviews | Average | Unhappy share |
|---|
Words and phrases that mark unhappy reviews
| Phrase | In unhappy reviews | In 4–5 star reviews |
|---|
Unhappy reviews with no reply, oldest first
| Date | Stars | Version | Review | Draft reply (edit before sending) |
|---|
The model, and what it leaves out
- Release drop compares each version’s average rating with the previous version that has enough reviews. Versions are ordered by their first review date, or by file order when there are no dates.
- Themes are keyword lists, so you can see why a review matched. A review can match several. The “overall average if fixed” column is a what-if: those reviews rated like the rest of the file.
- Phrases are single words and two-word phrases that appear in a larger share of unhappy reviews than of 4–5 star reviews, counted once per review.
- Reply queue needs a developer reply column; without one it is left out rather than guessed.
- Reconciles. Ratings, months and versions each add up to the same review count, and a test asserts it.
- Not modelled: meaning. Sarcasm, other languages, reviews that mention a bug in other words, and whether a theme actually caused a rating. Read the reviews behind any number you act on.
Common questions
How do I export my app reviews?
In Google Play Console open Quality, then Ratings and reviews, and use the download reviews option to get a CSV. For the App Store, Apple’s review pages do not offer a full download everywhere, so paste or export your reviews from App Store Connect or from a review-export tool you already use. Any CSV with a rating column and a review text column works.
What does the tool find?
Four things: which release made your rating drop, which complaint themes cost you the most stars, which words and phrases show up in unhappy reviews and not happy ones, and which low-star reviews you have not replied to.
How are themes detected?
With plain keyword lists, for example crash, freeze and force close for the crashes theme. It is simple on purpose so you can see why a review matched. It will miss some reviews and sometimes match the wrong one, so read the reviews behind any number you act on.
What does the star cost figure mean?
It is a what-if. If the reviews that mention a theme had rated like every other review in the file, the overall average would be that much higher. It is not a forecast and it does not prove the theme caused the low rating.
Is my review file uploaded?
No. The file is read in your browser tab and never leaves your device. There are no accounts or analytics vendors.
Is this advice?
No. It applies counting and averages to the file you provide.