1. Verdict
Why a ladder instead of one target
This idea is assessed at six MRR levels rather than one fixed target. A consumer subscription product's outcome is highly skewed: most reach little, some reach a comfortable monthly figure, and very few break out. A single "$20k MRR" number hides the most useful information: how likely each level is, and what each level costs in time.
The anchor: what happens to new subscription apps
RevenueCat's State of Subscription Apps 2026 (all categories, apps using its platform):
- 17.3% of newly launched apps reach $1k monthly revenue within 2 years.
- 4.6% reach $10k within 2 years.
- Median monthly revenue one year after launch is about $72.
- AI-powered apps earn 41% more per payer but churn 30% faster.
That base rate includes many hobby apps with no deliberate distribution. Astrology is a category where users already pay often. Both facts push the odds up for a well-executed plan. Fast AI churn and dominant incumbents push them down.
Odds at each MRR level (reached by month 24)
| MRR at month 24 | Base rate: new subscription apps | As described (palm + Vedic + Western + chat, everyone, native apps) | Palm-scan paid-ads funnel ¹ | Human-astrologer marketplace | AI tool for practising astrologers (B2B) | Recommended, ~15 hrs/week | Recommended, full-time |
|---|---|---|---|---|---|---|---|
| $1k | 17.3% | 20% | 25% | 10% | 35% | 45% | 60% |
| $3k | — | 10% | 15% | 5% | 20% | 30% | 42% |
| $5k | — | 6% | 11% | 3% | 12% | 22% | 33% |
| $10k | 4.6% | 3% | 7% | 2% | 6% | 12% | 20% |
| $20k | — | 1–2% | 4% | 1% | 2% | 6% | 11% |
| $50k | — | <1% | 2% | <1% | <1% | 2% | 4% |
¹ Needs about $3–5k a month in ad spend, full-time creative testing, and a funnel design that stays clear of the practices in the June 2026 FTC case. At this level MRR is not profit: ad spend can consume most of it.
Recommended means the calibrated Vedic AI companion described in WHAT-TO-DO-AND-AVOID.md, with the 3 conditions met: a Jyotish advisor, a sustained content routine, and 50 paying before building. If any one condition is missing, multiply the recommended odds by about 0.65.
These are judgment estimates anchored to the RevenueCat base rate, not computed statistics.
How to read the ladder
- $1k–3k is a realistic outcome with limited weekly hours, because the category already pays and the product runs without sales calls.
- $5k–10k needs retention to work. The calibration and memory features must measurably cut churn, and the diaspora USD share must grow.
- $20k and above almost always needs paid acquisition that pays back, or a breakout content presence. It's a different operating mode, and the odds show it.
- Where odds fall fastest: between $5k and $10k. That's the point where organic channels usually stop being enough.
Portfolio definitions (for comparison with other ideas)
| Tier | Definition | Recommended, ~15 hrs/week | Recommended, full-time | As described |
|---|---|---|---|---|
| Tier A (venture) | $1M ARR within 36 months | <1% | ~2% | <1% |
| Tier B (profitable business) | $20k+ MRR within 24 months, profitable, ≥80% gross margin | ~6% | ~11% | 1–2% |
These are lower than the B2B ideas in this portfolio at every level, for three reasons: consumer products have much lower base rates, AI apps churn faster, and the incumbents here are large and profitable. What this idea offers instead: first revenue within weeks, no sales process, a product that runs asynchronously, and a buyer who already spends on the category.
Fundable in current form?
No. AstroSage already runs an AI astrologer at scale, Astrotalk is valued at $1B, and several AI-first startups are funded and ahead. Nothing in the idea as described is defensible yet. See 06-fundability.md. It doesn't need funding to reach the levels where its odds are reasonable.
One-paragraph verdict
Astrology is one of the few consumer categories where people pay often, in small amounts, for personal answers. Astrotalk turned that into ₹1,176 Cr of revenue in FY25, 65% of it from marriage questions. "An AI chat about your kundli" is no longer unique: AstroSage has one, funded startups have them, and ChatGPT will read a pasted chart. The palm scan is a proven hook but a weak product, a biometric-data risk, and the signature feature of a funnel model the FTC is now suing. The part of your idea that is genuinely different is the calibration step: asking dated, discriminating questions about the user's past to rectify their birth time and prove accuracy before predicting anything, then remembering their life so every answer gets more specific. Build that for Indian and Indian-origin users, in Hindi, Hinglish and English, on web and WhatsApp, with no per-minute meter and strong safety rails. Then $1k–5k MRR is a reasonable expectation, $10k is achievable with good retention, and $20k+ is a stretch that needs paid growth.