Methodology
Charts. Once a day we read the Top Free and Top Grossing charts (top 100) in United States, United Kingdom, Canada, Australia, Germany, France, Japan, Brazil, India: overall, plus Productivity, Photo & Video, Education and Utilities/Tools. App Store data comes from Apple’s public iTunes RSS feeds; Google Play data comes from the public store pages.
What counts as an AI app. Apps whose main product is AI: a hand-curated list (ChatGPT, Gemini, Claude, Grok, Perplexity…) plus apps with AI in their name. We leave out general apps that only mention AI features.
Two leaderboards. App Store (iOS) and Google Play (Android) are ranked separately; their charts aren’t comparable.
Climbs. For every AI app on a main chart (overall Top Free or Top Grossing, 9 countries) we compare its rank now with its rank in an earlier real chart read: normally the read closest to 24 hours earlier, and for 7-day moves the read closest to a week earlier. On the first day of tracking we compare two reads from the same day and say so. Each app is listed once, under its strongest climb. NEW means the app entered that chart’s top 100. Nothing is estimated or back-filled.
Momentum (0–100). One score per app per store: for each country we take the app’s best main-chart move on the same log scale described below (dropping out counts as falling to #101), add those up, then add a bonus of 0.08 for every country it’s climbing in and subtract half that for every country it’s falling in. So climbing in seven countries beats one big jump in one. The total is squashed onto 0–100 (100 × (1 − e−total/0.8)) so scores stay comparable from day to day. A country counts as “climbing” if the app newly entered a top 100 there or its best move scores at least 0.03 (about +1 spot at #20 or +3 at #90). Boards are sorted by momentum.
Reason tags (at most two per app, in this order of priority):
· Breaking out: newly in a top 100 in 2+ countries.
· Broad climb: climbing in 4+ countries.
· Steady climber: best main-chart rank improved on each of the last 3 daily snapshots. This switches on once we have enough days.
· Hot in [country]: climbing in only one country, but with a big move (score ≥ 0.15, e.g. #49 → #40).
· Grossing up: climbing Top Grossing in 2+ countries, or one big Top Grossing move (score ≥ 0.1), a sign people are paying.
· Top 10: in the top 10 of a main chart right now.
Whole-store context. Ranks are positions among all apps on the chart, not just AI apps. Next to each app we name the highest-ranked household-name app it is ahead of on its best chart, e.g. “#34 overall · above Snapchat”, or two when they sit within 5 spots of each other (“above Instagram & TikTok”). Only a short, hand-checked list of brands counts (TikTok, Instagram, YouTube, Netflix, Uber, Amazon and similar, plus a few clearly mainstream local names such as LINE in Japan, Mercado Livre in Brazil and Flipkart in India, used only in their own country), matched by exact store id rather than by name. If none is within 40 spots below, we just show the rank.
Today’s summary. The bullets at the top are generated from fixed templates over the same data (broad climbs, chart entries, top-10 breakthroughs, big height-weighted jumps), ranked by importance, one fact per app. No AI writing, and nothing is added by hand.
Ordering. The number we show is plain spots gained. The order uses a score that cares about where a move lands: we compare ranks on a log scale, roughly “by what fraction did its rank improve”. Going from #49 to #34 (about 30% better) beats entering at #82, and #10 → #5 beats #100 → #80. Technically the score is ln((old + 10) ÷ (new + 10)); the +10 cushion stops a one-spot move like #2 → #1 from outranking everything, and a new entry counts as coming from #101.
Tracking started Oct 19, 2026. Past days before that aren’t available from the stores, so history grows one day at a time.