Understand negative reviews after an app update, from ChatGPT or Claude
Your app’s rating drops after a release: your AI reads App Store and Google Play reviews, isolates what changed and quotes users. No console access, no export.
Published on November 1, 2026
In short
- With Persyst’s Google Play and App Store Connect connectors, your assistant compares the rating before and after a release and groups negative reviews by topic.
- It quotes the reviews themselves, with their counts, instead of summarizing from memory.
- Product, support or leadership ask with no access to the Play Console or App Store Connect.
Since version 4.2, our rating is dropping. What’s wrong, on Android and iPhone?
On Android, version 4.2 averages 3.6 against 4.3 for 4.1; on iPhone, written reviews since its release on October 8 average 3.4 against 4.2 the month before. One shared topic dominates: sync failing after the update (24 Android reviews, 15 iPhone reviews, for example “my notes have been gone since yesterday”). On Android only, 11 reviews target the new navigation bar.
Sources consulted
- Google Play reviews · statistics by version
- App Store reviews · statistics by week
- Google Play and App Store reviews · search
The problem: the rating drops, but why?
A release ships and the rating goes from 4.3 to 3.8. The consoles show the curve and a few recent reviews. To understand, you have to read dozens of reviews across two stores, sort them by topic and check they are about the new version. Support often knows before product does, with no numbers to show it.
An AI assistant does this sorting in one conversation, as long as it can read the reviews.
What the assistant does, step by step
With Persyst’s Google Play and App Store Connect connectors, the question “What’s wrong since 4.2?” runs this reasoning:
- Measure the gap (
get_play_review_statsby version,get_app_store_review_statsby week): review count, average rating and star distribution, before and after. - Read the negative reviews of the period (
list_play_reviews,list_app_store_reviews, filtered on one and two stars) to spot the topics. - Count each topic with a targeted search (
search_play_reviews,search_app_store_reviews): “sync”, “login”, “navigation”, with several wordings. - Quote the most representative reviews, on each store.
The answer separates what both platforms share from what only affects one of them.
Asking the right question
- Name the version and its release date: the version is known for Google Play, not for the App Store.
- Ask for numbers and quotes: “how many reviews per topic, with two examples each”.
- Compare with a reference period: “against the previous four weeks”.
Three questions to try
- “Have reviews mentioning crashes increased since the update?”
- “On iPhone, does the drop affect every country or only France?”
- “Draft a standard reply for reviews about sync, for support to review.”
Setup
- An administrator connects Google Play and App Store Connect in Data sources: the Google Play and App Store Connect guides cover the steps.
- They add the apps, then open them to product, support or leadership.
- Each member adds Persyst to their assistant (getting started) and asks their question.
Limits to know
- Without the Play Console review reports address, Google only returns the last 7 days of reviews: set it to compare with an older period.
- The latest hours of reviews may be missing: they are refreshed every 6 hours.
- Reviews only: no crashes, installs or sales. A crash nobody reviewed does not show up.
Questions
Frequently asked questions
Do I need Play Console or App Store Connect access?
No. An administrator connects both accounts once in Persyst and opens the apps to the relevant teams. Members ask about reviews with no console account.
Why doesn’t the version show for iPhone reviews?
Apple does not send the app version with reviews. For the App Store, compare periods: before and after the release date, by week.
Does the computed rating match the stores?
Not exactly. Persyst computes over imported reviews: Google weights its rating differently, and Apple does not expose star-only ratings. Gaps between versions or periods remain comparable.
Can the assistant reply to unhappy users?
It can draft the replies, using what each review says. Publishing happens in the consoles, by someone with access: Persyst publishes nothing.