Repository navigation
win plan eoy2026
Companion to bear-case-eoy2026.md, written the same night. Rule of engagement: this plan may not contradict any sourced fact in the bear case — it must win inside them. Where the bear case says "ceiling," this plan changes the shape of the work so the ceiling moves. Same falsifiable style: every play has a number and a date.
The bear case's decisive point is Section 5: even flawless execution = ₹24–48K MRR by December. Correct — so revenue is the wrong 2026 scoreboard. A solo founder's end-of-2026 win is proof of three repeatabilities, in this order:
- A school pays and keeps paying (WTP is real, not polite),
- Teachers use it without you in the room (adoption defies the 85% base rate),
- School #2 arrives cheaper than school #1 (a channel exists).
Anyone can then finance/scale that in 2027 — via SISFS (₹20L grant + ₹50L, needs exactly this evidence, per your PROGRAMS.md), revenue, or a raise when the category thaws. Without those three, no amount of 2026 hustle matters. Every play below serves one of the three.
The bear case is right that ₹6,000/month is a visible cost to a ₹1.2L/month school. So never sell it as a cost.
- The audit (today) collects the number that matters: what the school loses to unpaid/late fees and what it already spends on diaries, SMS packs, or exam printing. In a 300-student school, if even 3% of annual fees go uncollected, that's ₹43,000/year (300 × ₹400 × 12 × 3%) — 7 months of Class-mode. Don't assert the 3%; measure it in their register during the audit and price against their number.
- Fee-season alignment: APS cash arrives when fees arrive. Offer annual billing at fee-collection time, monthly only as fallback. A ₹60–70K annual line item negotiated in fee season beats ₹6K/month extracted in the lean months.
- Floor discipline (bear-case failure mode #4): the printed offer is the price. The permitted concession is time (extend the unbilled setup window), never the per-student rate. A price cut you can't reverse; a free month you can.
Number & date: first invoice actually paid (money moved) by Aug 14 — the same day the bear case uses as its fail line.
The research says tools fail when they add steps (SchoolAI, Cambridge). The winning 15% subtract steps and keep a human champion. Concretely:
- Match the paper register exactly — the plan's design spec already binds this (status taxonomy mirrors the school's actual marks; 44px targets; vernacular). TD1 (language) gets answered today at the audit. The vernacular pass ships before the streak clock starts Sep 8.
- Appoint a champion, not a user base: one teacher, chosen with the principal, becomes the school's Class-mode person. You train the champion; the champion trains the rest. This is the only known antidote to "leadership commitment fades after rollout" — and it converts your service intensity (the bear case's ceiling) into a transferable asset.
- The principal gets a ritual, not a login: the daily absence summary and the term report land in the principal's hands (WhatsApp message you send manually at first — honest, labeled manual, per the approved plan). The principal's habit is what makes the teacher's habit stick. Dependence — your thesis's moat — is built at the principal's desk.
- Instrumented, not assumed: attendance_view/report_view events (shipped this week) make adoption measurable per teacher per day. The weekly metrics script is the adoption truth; zeros stay zeros.
Number & date: ≥60% of teachers marking ≥4 days/week with zero founder visits in the measurement week, by Sep 30.
The bear case prices founder-led cold sales at 1-in-5 and 2 schools/month. Both improve only with warmth and geography:
- Cluster rule: every next school within 30 minutes of school #1. One neighborhood = one support radius = one reputation pool. APS owners in a locality know each other (many belong to associations — NISA claims 400K+ member-adjacent schools nationally); a working register at one school is visible proof at the next.
- The referral ask is part of the S4 script: whatever today's answer, ask the principal: "which two principals should see this term report?" A referred visit converts at multiples of cold (and bear-case falsifier #3 — an unengineered referral by Oct 31 — becomes engineered-then-organic).
- The artifact does the selling: the printed term report (grayscale, letterhead-style, per the design spec) is a leave-behind that a principal shows other principals. Marketing budget: the cost of paper.
- S5 (candidate list) executes inside the cluster rule — 3–5 schools, all within the radius, intro paths via school #1's principal and the local association chapter.
Number & date: school #2 signed by Oct 31, sourced via school #1; CAC measured in founder-hours and printed pages, target <50% of school #1's.
- Aug–Sep: founder does everything, but writes the onboarding playbook as it happens (every step that worked at school #1, in order, in the school's language).
- Oct: school #2 is onboarded by the playbook + champion model, founder present but hands-off — the playbook is the product being tested.
- Nov–Dec: support moves to one weekly WhatsApp office-hour per school + the manual summaries automated (only after the streak evidence, per the demand-first rule).
- Cash mechanics run in parallel, none displacing school-facing hours (your 20% ops budget): Founders Hub ($1K Azure — kills the infra bill) this week; DPIIT after entity; SISFS application the week after the second school signs — a signed paid pilot plus adoption metrics is precisely the evidence that program funds, and ₹20L grant ≈ 3+ years of the bear case's infra-and-travel burn.
Number & date: by Dec 31, ≤1 founder school-day/week across all schools, measured from your own calendar.
- No AI as a selling point. Goldman Sachs' skeptics are directionally right for this buyer today; your own outage data (zero complaints) proves nobody's paying for generation. Gemini billing gets linked for the day-14 precondition, capped, and AI stays a quiet Lab feature gated on payment — per the approved thesis.
- No feature war with Teachmint. Free-and-funded wins feature wars. We sell the register + the principal's ritual + a human in the school. Depth in one workflow, not breadth.
- No expansion before the audit justifies it (binding sequencing guards). No second wedge, no parent app, no marketplace. The wedge is attendance→report→fees; everything else waits for 2027.
- No unpaid pilots. The bear case's failure mode #1 (polite yes, never pays) is refused contractually: the offer has a signature line and billing starts at marking, not at "interest."
| Proof | Metric | Bear-case base rate it defeats |
|---|---|---|
| WTP is real | ≥5 schools invoiced AND paid; ₹30K+ MRR collected (not booked) | verbal-yes trap; APS price sensitivity |
| Adoption is real | ≥60% teacher weekly-active across schools, founder absent | 85% poor-fit / 65% discontinuation |
| Channel is real | ≥2 schools sourced by referral; CAC falling school-over-school | no-brand cold-sales ceiling |
| Founder is free | ≤1 school-day/week total support; playbook executed by a champion at least once without founder | founder-becomes-the-product |
| Cash survives | infra ≤ credits (Founders Hub); SISFS application filed with pilot evidence | negative-margin hard job |
Hit all five and end-2026 Class-mode is a small, alive, provably repeatable business — which in this funding climate (worst edtech VC decade) is rarer and more fundable than a subsidized growth story. Miss them and the bear case was right — and your own escalation rule already says what to do next: mapped real buyer, second market, different wedge. Either way, nothing about 2026 is left vague enough to argue about.
The whole plan starts with one phone call at 8:30 this morning.