90-day validation

Six experiments, about ₹3.2 lakh. Ordered by uncertainty removed per rupee. Two of them are the whole plan.

E0 — Legal opinion on outreach · ₹40,000 · 1–2 weeks

Written opinion covering calls to published business numbers, DND-registered personal mobiles used as business lines, AI/synthetic voice disclosure, and WhatsApp template messaging. Runs first because it can invalidate an entire channel.

E1 — Will businesses pay? · ₹90,000 · weeks 2–12 ★

One field rep visits 200 directory-ready Whitefield businesses. Free listing plus 60 days of free Featured placement, every listing instrumented. At day 45, the actual numbers are presented and the sale asked for, with price randomised across ₹999 / ₹1,499 / ₹2,499.

Success: ≥15% trial-to-paid overall and ≥10% at the ₹2,499 cell. Failure: under 8%, or zero conversions at ₹2,499.

E2 — Will consumers contribute? · ₹60,000 · weeks 1–12 ★

Around 2,000 registered users, the 46 seed items live, weekly This/That pairs, AI community agent asking one contextual follow-up per human post.

Success: ≥4% contributor ratio, ≥30% of seeded questions receiving a human answer, ≥35% week-4 return. Failure: under 1.5% contribution, or under 10% of questions answered.

Report votes, comments and answers separately — a high vote rate with no answers is a failure, not a partial success.

E3 — What does a consumer cost? · ₹1,00,000 · weeks 3–10

Four parallel cells: Meta ads, ten nano creators, QR table-cards in twenty trial salons, and referral instrumentation. Success: any channel under ₹150 per registered user with ≥35% week-4 retention.

E4 — Does the graph rank? · ₹15,000 · weeks 1–12

Publish exactly 40 place pages, each carrying real community content. Not the 821. Success: ≥70% indexed by week 10 with impressions on at least 10 target queries. Failure: under 30% indexed, or any manual action.

E5 — Is Whitefield representative? · ₹15,000 · under a week

Run the existing pipeline on Indiranagar, Koramangala and HSR Layout. Converts the largest supply-side assumption into a counted figure. Cheapest item on the list — closes in week 1.

Decision gate at day 90

• E1 and E2 both succeed — raise a lean angel round against real conversion data.
• E1 succeeds, E2 fails — a directory and advertising business without community. Smaller, still real. Re-model without network effects or the intelligence layer.
• E2 succeeds, E1 fails — engaged community, unproven monetisation. Test lead-generation and commission models before raising.
• Both fail — don't raise. The thesis as constructed isn't supported.