Topical analysis

Topical analysis shows every topic your reviews have been classified into, for one app, viewed however you want to slice it. Where Insights tells the story of what’s currently unfolding, this page is the explorer.

Metric and view

A segmented control switches what the page is measuring:

• Volume: review count per topic (the default).
• Sentiment: average sentiment score per topic.
• Rating: average star rating of the reviews carrying that topic.

Alongside it, a view toggle switches how that metric is displayed:

• Treemap: every topic as a tile, sized and colored by the selected metric, with a topic-by-topic sparkline table beside it showing each topic’s recent trend at a glance.
• Table: a ranked list (Topic distribution) with review count, sentiment, average rating, share replied, and a trend arrow showing the change in review volume against the comparison period. Clicking a topic’s name opens its drilldown panel. Hovering a row shows two actions: AI summary for this topic opens the same panel, and Create an alert for this topic opens a new alert rule with the topic already filled in.

A category dropdown narrows either view to one topic category at a time, and a Manage topics link takes you to the topic taxonomy itself if you need to add, rename, or recategorize topics.

Narrowing by language

Because this is a single-app page, the shared language selector in the filter bar applies here too: pick a language to see the same treemap or table computed only from reviews in that language. This is how you’d check whether a topic is truly universal or concentrated in one locale. A spike in “Connectivity” that only shows up once you narrow to German is a very different finding than one spread evenly across every language.

Drilling into a topic

Clicking any topic (in the treemap, the sparkline table, or the distribution table) opens a drilldown panel scoped to that topic and the page’s current date range and language filter, with the sample reviews behind the numbers. This is the same drilldown panel used from Insights’ arc cards, so a topic you’re tracking looks and behaves the same way no matter where you opened it from. While the AI summary at the top of the panel is being written, a status line shows each step as it happens, such as reading the reviews, searching your help docs or older reviews for the same issue, and weighing what it found. A summary already written for that topic and date range opens straight away. The line above the summary says what it read, for example “Summary based on 376 reviews from Jul 16 to Oct 14”. The summary reads the topic’s last 90 days up to the end of your date range, so it can tell a new problem from a long-running one. The figures under it (reviews, rating, sentiment) count only your date range, which is printed beneath them, so they match the topic’s tile or row on this page.