We compared about 6 weeks before launch (Sep 15 to Oct 29) with the first 11 days after (Oct 30 to Nov 9), across all courses. Short version: the redesign is very good at getting new learners into a conversation with Rezzy, and does little yet for learners who were already here. The trial dip is real but comes from the onboarding paywall, not from Home.
My read: the new Home works as a front door. It turns new learners into Rezzy users at nearly twice the old rate. It hasn't yet changed how deep chats go or how often existing learners come back to Rezzy, and the generative invite card has barely been seen, so it can't be judged yet.
Share of each week's signups (with a course picked) who sent Rezzy at least one message within 3 days of signing up. New installs always get the latest build, so this is the cleanest before and after we have.
15.2% of 4,092 signups chatted within 3 days before launch; 25.4% of 751 after. The jump lands exactly in the release week and holds across most courses: US Medical PG 21% to 42%, CFA 8% to 28%, NCLEX 9% to 15%, MCAT 5% to 13%.
The new Home puts a Rezzy composer and a "first chat" card front and center, and the chat now opens as its own page. In the week of Nov 2, about 110 of roughly 154 learners who chatted opened Rezzy from Home. The old Home made Rezzy one option among many.
Indian Medical PG did not move (12.9% to 11.9%). These learners are mostly on Android, which was slower to update. Worth checking before we call it a universal win.
Rezzy chats come from the database, so they count every chat with at least one learner message. Learner counts are signed-in learners active in the iOS or Android app.
302 learners active both in Oct 1 to 26 and Nov 1 to 9, split by whether they updated to the new app.
| Group | Days with a chat | Msgs per active day | Msgs per chat day |
|---|---|---|---|
| Updated (231) | 36% → 33% | 1.87 → 1.66 | 5.2 → 5.1 |
| Still on old app (71) | 43% → 37% | 3.03 → 1.89 | 7.0 → 5.2 |
Both groups dipped in November, the updated group a little less. With only 71 learners on the old app, I wouldn't call that a lift. In the same days, 30% of updated learners chatted vs 21% on the old app, but the updated group is also more active overall.
Weekly chatters went from about 190 to 265, back to September levels. Most of that is more learners opening the app after the update (478 to 550 a week). The share who chat stayed at about 28%. For existing learners, the redesign hasn't yet made Rezzy a bigger habit.
Length is learner messages per chat. A "chat" is one Rezzy conversation with at least one learner message.
Average rose from 3.4 to 4.2 messages, but the median fell from 2 to 1 and the shape barely moved: 50% of chats are a single message, 34% reach three or more (32% before). The rise comes from a longer tail: 3% of chats now run 20+ messages.
Learner messages average 189 characters, down from 360. Learners are tapping starters and asking quick follow-ups rather than pasting long questions. That's healthy for engagement, but it means Home starts more conversations without making them go further.
46% of Rezzy chats created never get a learner message (49% before). Every chat page open that doesn't turn into a message is a missed first turn.
From RevenueCat, all platforms. Trials started after Oct 30 are mostly still running, so trial to paid conversion for them can't be read until around Nov 16.
Most trials start on the onboarding paywall, before a learner ever sees Home. On 1.69 and 1.70, the share of onboarding paywall viewers who tap purchase fell from about 71% to 53 to 57% on iOS. 1.69 also shipped App Review changes to that paywall (billed price first on every plan, trial charge date shown). Android went straight from 1.67 to 1.69, so it got all of this at once: trials from 1.67 fell from 19 to 4 a week as learners updated.
In 1.69, Home's paywall can't be closed by learners without access. In-app paywall views rose about 75% a week and purchases from them roughly tripled (2.5 to 8 a week). Direct paid subscriptions hit 22 in the week of Nov 2 vs 11 to 18 in October; revenue was $1,433 vs an October average of about $1,320.
Fewer trials now means fewer trial conversions in 1 to 2 weeks. At about 19% trial to paid, losing ~15 trials a week costs ~3 paid subscriptions a week, which the direct purchases currently cover. The hard paywall may also cost retention, and we can't see that yet.
Signed-in app learners, Oct 30 to Nov 9. 542 learners saw Home; 200 sent Rezzy a message.
The plain "ask Rezzy" box on Home is how most learners start (79 learners, 400 opens). Suggested starters, tool chips and daily prompts each reach only 13 to 27 learners, under 5% of Home viewers.
Only 9 learners saw the tutor's invite card and 8 tapped it (30 saw any Home card). It only appears after a learner has finished their first chat and chosen a study mode, which most haven't. So the generative piece has no measurable effect yet, good or bad.
Per-day averages before (45 days) vs after (11 days). Small courses swing a lot on a handful of learners, so read them as direction, not size.
| Course | Signups chatting in 3 days | Chatters per day | Messages per day | Msgs per chat | Chats with 3+ msgs | Chats after |
|---|
Prepared Nov 10, 2026.