AstroCompanion
As described:StopNarrowed:ContinueNo as described. Go as the calibrated Vedic companion.
astrology · ² At a blended ~$4.50 a month per payer (70% India, 30% USD). · last reviewed (next review by 28 Dec 2026) · 36 cited sources
Start this week
- Tell me your hours per week, whether you know a practising astrologer who'd advise, and whether you have any audience (Instagram, YouTube or WhatsApp) or can create content in Hindi.
- Choose the chart route (Swiss Ephemeris licence, or Skyfield + JPL), and run the 200-chart accuracy test.
- Draft the 12 calibration questions with the astrologer, and build the concierge form plus WhatsApp flow.
- Post the first 10 "find your real birth time" pieces, and take the first paid bookings.
Checkpoints and kill gates
Decide in advance what counts as on track. Missing a checkpoint means changing the buyer or the offer, not adding features.
Day 30
~15 hrs/week
50 paying for the concierge reading (₹299 / $9)
Full-time
50 paying for the concierge reading (₹299 / $9)
If you miss it
Change the entry question once (compatibility-first, or career-first). Missing twice means stop.
Day 90
~15 hrs/week
150 payers, ≥45% month-2 renewal, LLM cost <15% of revenue
Full-time
150 payers, ≥45% month-2 renewal, LLM cost <15% of revenue
If you miss it
Cut to the most-used flow (usually quiz → marriage/compatibility chat).
Month 6
~15 hrs/week
$1.5k MRR, ≥30% of new users from shares or referrals
Full-time
$1.5k MRR, ≥30% of new users from shares or referrals
If you miss it
Pivot A: diaspora-only, English, USD. Pivot B: rectification and matching as one-time paid reports (no chat subscription).
Month 12
~15 hrs/week
$4k MRR, ≥25% in USD, payer churn ≤10%/mo
Full-time
$4k MRR, ≥25% in USD, payer churn ≤10%/mo
If you miss it
Stop adding features. Fix churn and the share loop. Holding a profitable $3–4k is a valid outcome.
Month 18
~15 hrs/week
$7k MRR
Full-time
$7k MRR
If you miss it
Put all effort into the single best channel. Decide whether to test paid acquisition.
Month 24
~15 hrs/week
$10k MRR
Full-time
$10k MRR
If you miss it
Above $10k with CAC payback <3 months: consider buying growth for the $20k+ levels. $3–10k: run it profitably. Below $1.5k: stop, or pivot.
Do
1.Make the calibration quiz the front door. Ask 8–12 dated, discriminating past-event questions, rectify the birth time, and show what changed and why.
Target: ≥50% of signups finish it. Median "felt accurate" rating ≥4/5.
Why: It's the one part of the idea no competitor visibly does. It proves accuracy before predicting, and converts sceptics.
2.Get the chart exactly right. Tested ephemeris, Lahiri by default and shown, a vetted historical-timezone table for India, and the UTC offset shown.
Target: ≥99% match with a reference tool on 200+ charts
Why: Indian users cross-check. One wrong lagna or dasha date loses them.
3.Ground every chat answer in computed facts. The model calls the chart engine, and every answer cites the placement, dasha or transit.
Target: Zero invented positions in audits
Why: This is what beats ChatGPT and the inconsistent AI apps.
4.Build memory and consistency. Store the user's life context and one "reading" per topic, and answer from them.
Target: Same question, same core answer. Questions per payer ≥3 a week.
Why: Memory is the moat against the 30% faster churn of AI apps.
5.Lead with the questions people already pay for: marriage timing, compatibility, career and relocation
Target: Top entry topics = 70%+ of first questions
Why: 65% of Astrotalk's revenue is marriage questions.
6.Deliver the daily card on WhatsApp, personal to the user's transits and dasha, one message a day
Target: ≥3 opens a week among payers
Why: The daily habit already lives in WhatsApp.
7.Price simply and honestly. India: ₹199/mo or ₹1,499/yr plus ₹99 question packs. USD: $7.99/mo or $59/yr. Express consent, renewal reminders, one-tap cancellation.
Target: Refunds and chargebacks <3%. ≥40% of payers on annual plans by month 12.
Why: Honest billing is both the law and a trust signal.
8.Build the safety rails before launch: AI disclosure, crisis detection with helplines, 18+ only, and no medical, legal or financial instructions, death predictions or fear-based remedies
Target: Zero harmful transcripts in weekly audits
Why: California SB 243, New York §1700 and EU Article 50 already apply to chatbots like this. One screenshot can end the product.
9.Make outputs shareable: a compatibility result the partner must join to see, a "my real ascendant" card, and the daily card as a WhatsApp Status
Target: ≥30% of new users from shares or referrals by month 6
Why: The incumbents own search and outbid on ads. Shares are the channel you can own.
10.Add USD pricing for Indian-origin users abroad from month 7, English-first
Target: ≥25% of MRR in USD by month 12
Why: It's the incumbents' fastest-growing revenue line, and each payer is worth ~3x.
11.Launch on web (PWA) + WhatsApp. Consider the app stores only once the product is clearly "meaningfully different".
Target: Store submission only after month-3 retention is proven
Why: Apple's 4.3 rule names fortune telling. The web avoids the 15–30% fee.
12.Meter the model cost. Cache chart facts, and use a small model for routine answers.
Target: LLM and API cost <15% of revenue
Why: Heavy chat users on a flat plan can cost more than they pay.
Avoid
1.Making the palm scan the core, or claiming it predicts anything
Why: It's a 60-second hook with no evidence behind line reading. Palm imagery is biometric data under Washington law. Use it only as a labelled, process-and-delete extra, if at all.
2.Storing palm photos or hand templates
Why: Washington MHMDA (private right of action), Texas CUBI and Illinois BIPA exposure, for a feature that earns nothing.
3.Weekly plans and trials that silently convert
Why: The FTC sued the largest palm-and-horoscope funnel operator in June 2026. India's CCPA, California's ARL and the UK DMCC rules all target this.
4."Today is important: do everything by astrology" pushes
Why: Nudging high-stakes decisions creates liability and destroys trust when they go wrong. Offer favourable windows only for things the user says they're planning.
5.Answering health, legal, money or death questions directly
Why: Harm and liability. Reframe as timing and wellbeing, and route crises to helplines.
6.Vedic + Western + tarot + numerology at launch
Why: Different audiences and vocabularies. You'll be average at all of them.
7.A human-astrologer marketplace
Why: Ops-heavy supply, payouts and moderation, and Astrotalk owns it at $1B valuation. Refer or partner later if users demand it.
8.Remedy, gemstone and puja e-commerce
Why: Fear-based remedy selling is the category's trust problem, and it brings inventory and advertising-law risk.
9.Guarantee language in ads ("100% accurate", "guaranteed marriage")
Why: ASCI upholds complaints against exactly this.
10.AGPL chart libraries in a closed server app
Why: Either buy the Swiss Ephemeris licence (CHF 700, one-time) or build on Skyfield + public-domain JPL data.
11.Head-term SEO ("kundli", "horoscope") and English Meta ads against the palm-scan apps
Why: You can't outrank AstroSage or outbid funnel apps. Win the long tail and the share loops.
12.Native apps before month-3 retention is proven
Why: App review risk, fees, and months of work that don't tell you whether people pay twice.
13.Users under 18
Why: Parental consent under DPDP, and extra duties under SB 243. Not worth it.
14.Adding features when a checkpoint fails
Why: Change the entry question or the audience, not the feature list.