Confidential Joyed.ai Feb 17, 2026
— WEEK 1 + WEEK 2 —

EARLY
PMF
SIGNALS

First real user data for Joyed.ai web app.

6 students · 7th grade · Estonia Snapchat study group 2 homework quests Major test prep Zero paid acquisition
Retention Quest 1→2
67%
4 of 6 students who opened Quest 1 came back for Quest 2
Meaningful Engagement
67%
4 of 6 studied >1 min in Quest 1; all 4 in Quest 2
Avg Study Time / Quest
~18 MIN
Avg per engaged student, Quest 1 (Estonian)
PM + AM Sessions
3/4
PM study + AM revision pattern, Quest 1
01

What happened

Week 1 · Thu Feb 12 · Estonian Language Quest
First group quest sent via Snapchat
Quest link (homework assignment) sent to the Snapchat study group the day before a major Estonian language exam. All 6 students opened it. 4 engaged meaningfully — the longest session was 40 minutes across PM + AM. 3 of 4 returned AM before the test for a revision. 2 bounced in under 30 seconds. Founder sent a "wishing you luck!" nudge via Snapchat 10 min before the test.
Week 2 · Mon Feb 16 · History Quest
Second quest, same group
Second quest link sent the day before a major History exam. 4 of 6 students opened and engaged — zero bounces. No luck nudge was sent this time. Fewer students returned AM for revision compared to Quest 1, consistent with the nudge driving that behaviour. One student who bounced on Quest 1 came back for 7 minutes PM and returned AM — converted by peer social proof.
02

Engagement by student

Sofia
50 min total
Laura
26 min
Loore
12 min (Quest 1 only)
Aksi
9.5 min
Anz
7 min
Adeele
30 sec (bounced)
03

Quest vs Quest

Quest 1
Estonian Language
Students who opened6
Meaningful engagement (>1 min)4 of 6
Total study time71 min
Avg per engaged student~18 min
PM + AM session users3 of 4
Bounces (≤30s)2
Quest 2
History
Students who opened4
Meaningful engagement (>1 min)4 of 4
Total study time33 min
Avg per engaged student~8 min
PM + AM session users1 of 4 (Anz)
Bounces (≤30s)0

Note: Lower avg in Q2 likely reflects content depth (history vs language exam prep) rather than engagement drop. Zero bounces in Q2 is a positive signal — students who returned knew what to expect.

04

User retention

Student Quest 1 · Estonian Quest 2 · History Status
Sofia
40 min — power user
10 min
Retained
Laura
15 min
11 min
Retained
Aksi
4.5 min
5 min
Retained
Anz
30 sec (bounce)
7 min + AM revision
Converted
Loore
12 min
Did not open
Churned
Adeele
30 sec (bounce)
Did not open
Disengaged
05

Behavioral patterns worth noting

PM Study + AM Revision
A Natural Study Ritual
3 out of 4 engaged students on Quest 1 did a PM study session then returned for a quick AM revision before the test. Nobody was instructed to do this.
Bounce → Full Engagement
30 sec → 7 min Conversion
One student bounced on Quest 1 (30 seconds) but returned on Quest 2 for 7 full minutes PM — then came back AM for a revision. Peer social proof within the group drove re-engagement without any direct prompting.
Session timing
High-Intent Windows Only
Sessions cluster tightly around test windows: PM 17:00–23:12 and AM 07:00–09:48. Aksi studied at 11pm. This is deliberate, high-stakes study behavior — not casual browsing.
Nudge effect
AM Revision Is Nudge-Driven
Quest 1 (week 1): founder sent "wishing you luck!" to the group 10 min before the test — 3 of 4 students returned AM for a revision. Quest 2 (week 2): no nudge sent — 1 of 4 returned AM. Same group, same product, 3× difference. The nudge mechanic validated itself across the first two weeks.
06

The signal

First launch: up to 40 min study sessions.
We built something teenagers needed.
67% returned for a second subject —
they preferred Joyed.ai over studying alone.
The product is fitting a real study habit:
PM deep, AM fast.

Each "quest" is a homework assignment sent as a link to the Snapchat study group the day before a major test. Students click, study, and close — session IDs connect their data across quests without requiring an account. On Quest 1 (week 1), a simple "wishing you luck!" nudge sent 10 minutes before the test brought 3 of 4 students back for an AM revision. On Quest 2 (week 2) with no nudge, only 1 returned. Same group, same product — 3× difference. The nudge mechanic validated itself in the first two weeks.

07

What students said

Outcome validation "

Estonian test felt easy, I was the first one to hand it in in the classroom. I think your app is part of the reason.

Sofia · Quest 1 · 40 min session
Product signal "

I forgot to do the quick review the second time, before the History test. I'd like the app to always send notifications 30 min before the test, to remind me about the 1-minute-summary.

Aksi · Quest 2 · 5 min session
Joyed.ai · Early PMF Report · Feb 2026 n=6 students · Weeks 1–2 · Biweekly · Confidential