AstroCompanion

This pre-mortem covers the idea as described (palm + charts + chat + daily guidance, for everyone). Causes are ranked by probability.

---

Cause #1: The novelty wore off, and nobody paid for month 2 (probability ~40% of failure cases)

What happened. The palm scan and the first kundli reading impressed people. Signups were fine. Users asked the bot 5–10 questions in the first week, got answers that sounded like every other horoscope, and stopped opening the app. Paid users cancelled after the first month: AI apps churn about 30% faster than other subscription apps (RevenueCat 2026). The free alternatives were one tap away: AstroSage's free kundli, the free first consultation on human-astrologer apps, and ChatGPT with a pasted chart.

Early warning signs

• Day-7 return rate below 20%, day-30 below 10%.
• Median chat questions per user per week falls below 2 after week 2.
• Monthly-plan renewal into month 2 below 50%.
• Users say "it's generic" or "it's like my horoscope in the newspaper".

Fixable? Yes, and the fix is already in your idea. The calibration quiz, memory and personal daily card are exactly what makes month 2 different from week 1. The failure is shipping the chat without them.

---

Cause #2: Couldn't reach users at a cost that paid back (probability ~35%)

What happened. Search results for "kundli", "horoscope" and "kundli matching" are owned by apps with 15+ years of domain authority. Meta and Google ads for astrology cost more than a ₹199 subscription earns before it churns. The big Western palm-scan apps outbid everyone on English-language ads. Without an audience, growth stalled below a few hundred payers.

Early warning signs

• Paid CAC above 3 months of ARPU.
• Organic signups flat after month 4.
• Under 20% of new users arrive through shares, referrals or content.
• All growth comes from one channel you don't control.

Fixable? Partly. Make every output shareable (compatibility results, calibrated "your real ascendant" cards, daily card on WhatsApp Status), build long-tail content that the incumbents don't bother with (specific transits, dasha periods and life events), and price for USD-paying diaspora users so each payer is worth more. See 07.

---

Cause #3: A trust, safety or platform event (probability ~25%)

What happened. One or more of:

• a user found the app's ascendant or dasha dates didn't match their family astrologer's or AstroSage's, posted it, and the reviews followed;
• the bot told a user in distress something harmful (to stop medication, to leave a marriage, that a death was coming), and the screenshot went viral;
• the "important day" nudges pushed users into financial decisions that went badly;
• the App Store rejected the app as a saturated fortune-telling category, after the launch was planned around it;
• stored palm photos created a biometric-data complaint, or the free-trial flow was reported as a subscription trap.

Early warning signs

• "Wrong chart" complaints above 1% of users.
• Any chat transcript in which the bot gives medical, legal, financial or fatalistic advice.
• App review feedback citing guideline 4.3.
• Chargebacks or refund requests above 3% of payments.

Fixable? Yes, by design:

• test the chart engine against a reference tool on hundreds of charts before launch, and show which ayanamsa and settings are used;
• put hard guardrails in the chat (crisis routing, no medical, legal or financial instructions, no death or disaster predictions, no fear-based remedy selling);
• frame important days as timing windows for things the user already plans;
• launch on the web, not the stores;
• process palm photos without storing them;
• use plain, one-tap cancellation.

---

Honourable mentions

• LLM cost creep. Chat-heavy users on a flat plan can cost more than they pay. Meter questions, cache the chart facts, and use a small model for routine answers.
• The competitor copies the quiz. Well-funded human-astrologer apps can add a calibration quiz in a sprint. Your defence is memory depth and how early you collected the data, not the idea.
• Credibility gap. Without someone who knows Jyotish well reviewing the interpretation rules, serious users will spot errors that a developer can't see.