Skip to content
Rankora
5 credits per run

App Review Trend Analysis

See whether an app's rating is rising or falling and which words keep coming back in the complaints and the praise of the latest reviews.

Privacy and cost

This tool uses paid live data. A run costs 5 credits, only charged when it succeeds. Recent identical requests are served from a cache, which keeps costs and prices low.

No subscription required for credits.

How it works

  1. Choose the store, the app, the country and the number of reviews to fetch (50 to 200).
  2. The tool fetches the most recent reviews, computes the average rating per week or month and compares the older half with the recent half.
  3. It counts words and word pairs once per review: terms in 1-2 star reviews, in 4-5 star reviews and terms whose share is growing in recent reviews.

Example

On 150 reviews of an iOS app in the French storefront you get a weekly rating curve, a verdict (improving, stable or getting worse) and the most frequent terms of 1-2 star reviews, such as the name of a feature that causes trouble.

What the result means

The analysis covers a recent sample, not the full history. "Sentiment" is simply the star rating of the review where the term appears: no AI, no hidden interpretation. A term needs at least 2 occurrences to be listed and a change under 0.2 points counts as stable.

Common problems this tool helps with

  • Understanding whether a recent update lowered satisfaction.
  • Finding topics that recur in a competitor's negative reviews, to use as a listing argument.
  • Spotting an emerging problem: a term whose share climbs in the latest reviews.
  • Feeding your description with users' real vocabulary.

Frequently asked questions

Do you use AI for sentiment?

No. Sentiment is the star rating of each review. Terms are plain counts of words and word pairs, with stop words removed.

How many reviews are analysed?

50 to 200, the most recent ones. For a very popular app that may cover only a few days.

Why a "not enough reviews" error?

At least 10 recent reviews are needed to compute trends. Some countries publish almost none.