App store intelligence splits into a board and a history. The board is get_top_trends with type App Store Top Free, App Store Top Paid, or Google Play. No keyword. History is get_growth or get_time_series with source app downloads or app rankings and an Android bundle ID. POST https://api.trendsapi.ai/api. Mode names sit on the API reference. Store-facing notes sit on the App Store guide. Unwrap body before reading a board or a history row.
Charts versus history
POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json
{"mode":"get_growth","source":"app downloads","keyword":"com.openai.chatgpt","percent_growth":["3M","12M"]}
A display name in keyword fails. An iOS App Store ID fails. If the POST also lists other sources, the keyword must still be the bundle ID.
ChatGPT downloads fell on the quarter and rose on the year
Bundle-ID history from 27 Aug 2026, com.openai.chatgpt on app downloads: recent value 10.0 on 26 Aug 2026. 3M growth −79.17% (10.0 vs 48.0), recent volume 7,293,948 vs 34,826,884. 12M growth +63.93% (10.0 vs 6.1), baseline volume 4,404,639. A dashboard that only plots the year calls the app a winner. The quarter is a crash from a spring peak.
app rankings for the same bundle ID returned date_out_of_range on 3M and 12M. Available points in that pull ran 29 Jul 2026 to 27 Aug 2026. A later short-window retry returned no_data. Treat ranking history as a short tape. Do not invent a rank percent.
The Google Play board is names, not IDs
type Google Play, limit 6, snapshot 2026-08-27T18:01:38Z: WhatsApp Messenger, Facebook, YouTube, Instagram, Free Fire: 9th Anniversary, Messenger. Those strings are chart titles. They are not valid app downloads keywords. Sensor Tower Connect still sells panel usage and Snowflake feeds for teams that need estimated DAU. This product does not reprint that panel. It reprints the public board and a bundle-ID series.