Product Bet Hypothesis: Audio “Podcast Feed” vs Text/Video “Feed”

Hypothesis

One of these is the true “primary value experience” for the beachhead:

Bet A (Audio-first): Learners pay to get a personalized podcast feed (Pimsleur-like audio lessons + conversation podcasts) generated from content they care about.
Bet B (Feed-first): Learners pay to get a LanguageReactor-like feed of target-language content with comprehension support (reading/listening UX).

Learners often resist “homework” and prefer low-friction consumption; audio can fit commutes/chores.
Many tutors already assign audio/video media (films, series, street Portuguese) as authentic practice.

Audio practice signal:
- “Dou atividades mais práticas: mando áudio, a pessoa terá que descrever o que entendeu; assim a pessoa habitua-se ao sotaque e perde a vergonha.” — “I give more practical activities: I send audio; the person has to describe what they understood; that way the person gets used to the accent and loses their shyness.”
Feed/content signal (media recommendations):
- “Eu indico filme e seriado com gírias; muitos querem aprender o português das ruas, principalmente do Rio de Janeiro.” — “I recommend films and series with slang; many want to learn ‘street Portuguese,’ especially from Rio de Janeiro.”
Constraint (extra tooling resistance):
- “Todos os alunos que eu perguntei quer que eu mande alguma coisa? Não… Não gosta. Só quer conversar, só quer falar.” — “Every student I asked, ‘Do you want me to send you something?’ No… they don’t like it. They just want to chat, just want to speak.”

If the “feed” experience requires attention and sustained time, it may fail the “busy adult” constraint.
If audio-only doesn’t create visible progress, learners may churn.

Bet A: user can generate + play a tailored audio lesson quickly, and it feels “made for me.”
Bet B: user can consume content with support (subtitles/definitions/highlights) immediately with minimal setup.

Time-to-first-value: minutes from signup to first successful “moment” (play audio / consume supported content).
Week 1 habit formation: days active; minutes listened/read; return rate within 48 hours.
WTP: explicit willingness to pay at $12/mo; “choose this over Duolingo/LanguageReactor/Pimsleur” statements.
Retention: week-4 retention (later, after initial activation is solved).

Concept test interviews (no build needed): show two one-page mocks; ask which they’d buy and why.
Artifact test with the current app:
- Run 5 learners through “upload → generate Pimsleur-like audio lesson”.
- Run 5 learners through “content feed” concept using a lightweight prototype/mocks (until implemented).
Ask a forced-choice: “If we only shipped one for 3 months, which would you pick?”

Double down on Audio-first if:
- ≥60% of learners prefer audio-first and can name a routine where they’d use it ≥3 days/week.
- Activation is higher for audio: ≥50% reach first value in ≤5 minutes (in a guided test).
Double down on Feed-first if:
- ≥60% prefer feed-first and describe using it as a “default” daily habit.
Hybrid if preferences split but both show strong WTP; start with the one that yields higher activation.
Kill if neither bet produces WTP or realistic weekly routines.