Doc 1 · owner: Rebekah · drafted 2026-08-06 · status: draft for Isaac's review
The Plan
Goal
Answer two questions before the September consumer launch:
- Engagement — do students keep using the app without a teacher or parent pushing them?
- Improvement — does 2–3 weeks of daily use measurably improve spoken English?
Secondary: surface friction (onboarding, mic permissions, payment attitudes) and measure willingness-to-pay via the day-21 continuation offer (RM15/month).
Participants
- Target N = 100 recruited (plan for up to ~50% dropout → ≥50 completing). Raised from 24–30 — exceeds the 10–50 range from the Isaac sync, so flag the new scale to Isaac. Personal contact runs on a sampled model.
- Skew weaker English speakers — B/C English grades or teacher-nominated "quiet in English class". Strong students can't show improvement in 3 weeks; weak ones can.
- Form mix: from Chung Hua, recruit Form 1 and Form 5 specifically — the same cohorts as the 500-student trial (250 F1 + 250 F5), so results transfer directly. Students from schools 2–4 skew Form 4–5.
- Requirements: phone access, parent consent, and upfront agreement to the two short speaking assessments — joining means doing both.
Recruitment channels
| Channel | Owner | Next action |
|---|---|---|
| 1. Chung Hua (Seremban) — main group, target 60–70 students | Eugene | Confirm the school can go beyond the earlier 20–30 + school's blessing this week; still a direct dry run for the 500-student trial |
| 2. Audrey Kho — Catholic High School — up to 10 students | Rebekah | Contact received from Jon; message Audrey this week with the teacher pitch |
| 3. Nancy (via Jon) — teacher in Melaka — up to 10 students | Rebekah | Contact received from Jon; message Nancy this week |
| 4. Natasha Razali — SMK Putrajaya — up to 10 students | Rebekah | Contact directly (known from the earlier school interviews) |
Chung Hua fills the bulk. Schools 2–4 contribute up to 10 students each (the teacher pitch already asks for 8–10 nominees), for two reasons: diversity — Chung Hua is one school and Chinese-medium; schools 2–4 are SMK-side, so they stop the findings from being single-school artifacts and add a non-Chinese-medium comparison — and backup, if Chung Hua confirms fewer than hoped. All three contacts are teachers, so the same pitch goes to every one of them.
Recruit more than needed; confirm up to ~100 who return signed consent. If Chung Hua can't go past ~30, hitting 100 needs more schools — decide early in recruitment week.
Site logistics
The speaking assessments run inside the app (see The Measure), so kickoff sessions are onboarding only — install, account, mic check, walkthrough:
- Chung Hua (Seremban): Eugene runs it — at 60–70 students, split into two or three classroom-sized batches (or enlist a teacher); onboarding is only install + account + mic check, so each batch is ~45 min.
- Catholic High (PJ) + SMK Putrajaya: Rebekah attends in person — 2–3 students each, one short visit per school.
- Melaka (Nancy): via video call.
Students do the pre Speaking Check in the app that evening at home (quiet room); Day 1 unlocks after it. One WhatsApp group per cohort; same check-ins, same assessments everywhere.
The teacher pitch
Goes to teachers — they recruit the students. Kept in B1 English so it can be forwarded to students as-is; translate to BM/Chinese as needed.
Hi [Teacher's name], we are going to launch Kelas Sekejap for high school students. It's a web app that helps students speak English with more confidence.
We are looking for students to try it for 3 weeks and tell us what they think. The ask: practice is 10 minutes a day on their phone. We have rewards and prizes at the end of the trial!
Could you help us pick 8–10 students? It works best for students who are quiet or less confident in English — they do not need to be top students.
Each student will also do a short speaking activity in the app at the start and at the end (about 5 minutes each). It is not an exam — it helps us see how much they improve. Students who join need to do both.
Interested? Reply here and I will send a short info letter for parents.
This pitch is fully honest as written: the trial is free, and rewards/prizes at the end are the actual deal. The parent letter mentions the optional day-21 continuation offer so it never reads as a surprise upsell.
Incentive model
Agreed with Isaac · 2026-08-06
- Free 21-day trial — the whole study is free. No payment ask during the study, so the before/after measure keeps its N (the proof investors want comes first).
- Complete it → reward. "Complete" = practiced ≥15 of 21 days AND did the post Speaking Check + exit form. Reward: completion certificate + ~RM10 Grab voucher (≈RM500–600 for ~50–60 completers).
- Day-21 offer: continue at RM15/month — 50% off for the first 3 months (then RM30). Made at the exit session with the student's own in-app stats in hand (days practiced, WPM trend). This is the conversion measurement AND a preview of how a progress report sells the product at Chung Hua scale. Parent pays via TnG/Grab — payment friction gets tested here.
- Everyone signs the light commitment at kickoff: practice ≥5 days/week, answer the weekly check-in, do the speaking activity at start and end (already disclosed at recruitment — no surprises).
- Pending budget approval: bi-weekly cafe meetup for group speaking practice (2 sessions) — realistically Seremban-cohort only.
Schedule
| Dates | What |
|---|---|
| Aug 7–15 | Recruit, collect parent consent, WhatsApp group, pre-work checklist below |
| Sun Aug 16 or Mon Aug 17 | Kickoff (per site — see logistics): install, activate accounts, mic check, walkthrough. That evening at home: in-app pre Speaking Check (Day 1 unlocks after it) |
| Aug 17 – Sep 6 | 3 weeks of daily use (app days 1–21). Weekly: 5-question check-in (all students) + rotating sample of personal calls (~8–10/week, split Rebekah/Eugene) |
| ~Aug 29 | Optional cafe meetup #1 (if budget approved) |
| Sep 7–11 | In-app post Speaking Check (unlocks after Day 21) + exit form for every completer + 15-min interviews with a ~20-student sample → reward + the RM15/month offer to every completer's parent with the student's stats |
| By Sep 11 | Findings write-up → feeds launch decisions |
What we measure
Engagement (mostly already instrumented)
- Backend truth: days active per week, lessons completed, D7/D14/D21 retention — from per-account session data. (The streak counter is excluded: it's device-local.)
- PostHog: funnel drop-off within a lesson, session replays for friction moments (inputs masked).
- Completion rate (≥15 of 21 days + post-assessment) and day-21 conversion: how many parents take the RM15/month offer, and the stated reason of every "no" — the monetization result Isaac wants.
English improvement — the proof investors want
Full instrument in The Measure: an in-app Speaking Check (pre gates Day 1, post unlocks after Day 21 — no human administration, which is also what lets it run at 500-student trial scale), counterbalanced A/B speaking tasks, a model-scored rubric (/15: fluency · vocabulary · grammar — all transcript-gaugeable) validated by a blind human sample, objective transcript metrics, and pre/post self-rated confidence captured in the app. Output: a one-pager per cohort plus 3–5 paired before/after audio clips (parent consent checkbox required).
Honest caveat for the write-up: 3 weeks is short; expect movement in fluency and confidence, not grammar. Frame results as directional evidence the 500-student trial confirms.
Qualitative
Weekly check-in (5 questions), one call per student, exit interview. Standing questions: what made you open the app / skip a day; favourite and most confusing part; would you pay RM30/month — would your parents?
Pre-work checklist
- Confirm prod build + accounts flow for 30 external users (activation codes?)
- Verify PostHog key live in the build they'll use; test one end-to-end session
- Decide: listening module on or off for this group (content done; audio generation hasn't run — likely off)
- Build the in-app Speaking Check — the one engineering deliverable; tested end-to-end on a real phone by Aug 16
- Rest of assessment prep (final A/B prompt copy, model-scoring script, admin spreadsheet)
- ✓Parent letter drafted (EN/BM/中文, all disclosures + optional audio-clip checkbox). Remaining: contact number, reward wording after budget yes, paper vs Google Form per school
- WhatsApp group + weekly check-in form created
- Payment path for the day-21 offer: TnG/Grab collection a parent can complete (manual TnG QR fine if web checkout isn't live)
- Budget call from Isaac: cafe meetups (~RM300 × 2) + completion rewards (~RM600) + prints ≈ RM1,300 total — and flag that N=100 doubles the originally discussed 10–50 group size
Risks
- Chung Hua concentration → 60–70 of ~100 students come through one channel; if Eugene or the school stalls, the group collapses. Get the school's yes in week one, and keep schools 2–4 warm as fallback, not just diversity.
- Exam-season collision → confirm dates before kickoff; fallback is a holiday-period "camp" framing (mirrors the Chung Hua summer-camp option).
- Dropout → recruit to ~100 assuming up to 50% loss; commitment contract, weekly check-ins, sampled personal calls.
- Weak students stall at onboarding → onboarding done together at kickoff, BM/Chinese support in the WhatsApp group.
- Improvement signal too noisy → three instruments (blind-rated task, attempts-data trajectory, self-report); report convergence, not one number.