NST Founder Lab · Week 8

Answer Key — Instructor Only

Enter the password to view the debrief key.

ANSWER KEY / DEBRIEF

This key gives defensible reasoning, not a single mandatory investment verdict. Reward explicit assumptions, denominator discipline, evidence relevance and the ability to change a decision when evidence changes.

1 — CLAIM → EVIDENCE → GAP

ClaimEvidence presentGap / analytical interpretation
Pain is real and recurring31/42 report weekly missed calls; 23 can show logs; 28 use workaroundDiscovery sample selection is unknown; frequency/economic loss per restaurant not quantified
Customers receive useful value12 installed; 8 active D14; 5 paid; 3 renewed; one asked for second outletVery small cohort; renewal depth immature; which feature creates value is unclear
Acquisition can scale economically9 paying; 6 attributable to paid/outbound; spend/tools data suppliedFounder time excluded; one city; repeatability and channel saturation unknown
Large scalable market exists85k profile-matched estimate; 60% compatibility; 12% reachable assumptionConversion across reachable population unproven; adjacency claims unsupported

2 — SAID → DID

Useful chain: 26 said try → 18 agreed pilot (69.2%) → 12 installed (66.7% of agreed; 46.2% of said-try) → 8 active D14 (66.7% of installed) → 5 paid (62.5% of D14 active; 19.2% of said-try) → 3 renewed (60% of paid; 11.5% of said-try).

Strong signals: visible missed-call logs/workarounds; actual installation; payment; renewal; second-outlet request.

Weaker signals: 'would try' statements; pilot willingness; percentages from tiny denominators.

A strong answer identifies where belief changes, rather than declaring one conversion 'good' or 'bad' without context.

3 — VALUE PROPOSITION CANVAS

Likely initial segment: Independent Casual Dining is defensible because pain is high, pilot willingness is highest, integration is low and owners are directly reachable. Students may defend another segment if evidence-based.

Jobs: answer high-intent customer calls during peak service; capture reservations/basic enquiries/orders without pulling staff away.

Pains: missed calls; staff distraction; manual/personal-phone workaround; discomfort with AI handling unusual requests.

Gains: fewer missed high-intent calls, reliable basic handling, staff escalation for complexity, potentially more captured reservations/orders.

Pain relievers/gain creators must map to supplied product performance; unsupported future features should be penalized.

Example value proposition: 'For independent casual-dining restaurants losing high-intent calls during peak service, VoxKitchen handles routine inbound requests and escalates complex calls so staff can serve in-restaurant customers without simply letting the phone ring.'

4 — CONTRADICTION HUNT

96% answered vs only 61% fully resolved: answering is not the same as successful resolution.

Strong stated willingness (26 would try) vs only 5 paid and 3 renewed: intention collapses through the funnel.

Order capture is part of the thesis, but two restaurants disabled it; order-intent accuracy is 78%, and multilingual success outside Hindi-English is 58%.

Management talks multi-city scale, but acquisition evidence is founder-heavy and from one city.

5 — ASSUMPTION CHAIN

A strong chain may be: recurring missed-call pain [M/E] → target restaurant has sufficient call volume [E] → owner reachable [M/E] → agrees pilot [M] → installs [M] → AI resolves valuable calls [M/E] → customer perceives economic/operational value [E] → pays [M] → renews [M, tiny n] → channel CAC remains viable without founder heroics [G/E] → city-to-city repeatability [G] → 250 restaurants [G].

High-impact weak links include retention/value delivery and scalable acquisition. Accept either if defended from case evidence.

6 — SELECTIVE QUANTITATIVE AUDIT

CAC claim:

₹45,000 ads + ₹36,000 tools/agency = ₹81,000.

₹81,000 / 9 total paying = ₹9,000. This appears to be management's blended-style calculation.

But only 6 paying customers are attributed to paid/outbound. Using the supplied acquisition spend against those 6 gives ₹13,500 per paid/outbound customer, before valuing 150 founder hours.

Founder time is economically real even when cash accounting records it at ₹0.

LTV claim:

₹4,600 × 12 months = ₹55,200, so management appears to use revenue × assumed lifetime.

Monthly service cost: 510×₹3.20 + ₹550 + ₹300 = ₹2,482.

Approx. monthly contribution before fixed overhead = ₹4,600 − ₹2,482 = ₹2,118.

Contribution-based 12-month scenario = ₹25,416. A 6-month scenario = ₹12,708.

The central weakness is not arithmetic; the company is ~10 weeks old, so a 12-month lifetime is not observed.

Runway claim:

Current monthly operating cost = ₹4.2L + ₹1.1L + ₹0.9L + ₹0.3L = ₹6.5L.

Current revenue = ₹0.414L, so approximate current net burn = ₹6.086L/month.

Current cash runway ≈ ₹11L / ₹6.086L = 1.81 months.

Post-raise cash ≈ ₹86L. At unchanged burn, ≈14.13 months.

If costs rise by ₹2L/month after Month 3 and revenue is held static for a conservative simplification: first 3 months burn ≈₹18.258L, remaining cash ≈₹67.742L; later burn ≈₹8.086L/month, giving ≈8.38 more months; total ≈11.38 months.

Therefore 'approximately 12 months' is plausible under this simplified cost step-up, but highly dependent on revenue growth, hiring timing, working capital and other cash costs.

7 — 96% CLAIM

96% means the AI answered the incoming call, not that it autonomously solved the customer's job. Only 61% are fully resolved without human intervention; 35% transfer; order-intent accuracy is 78%; other-language success is 58%. A defensible IC rewrite: 'The system answers 96% of inbound calls in the current sample; 61% are fully resolved without staff intervention, with weaker performance in order capture and non-Hindi-English languages.'

8 — DECISION-CRITICAL UNKNOWN

Best answers attack a high-impact weak link. Example: retention/value.

Unknown: whether restaurants receive enough recurring value to renew after novelty/pilot support fades.

Experiment: recruit a defined cohort of qualified independent casual-dining restaurants, run 30 days with normal onboarding/support, charge a real price, and track usage/resolution/renewal.

A defensible student threshold might be ≥20 qualified pilots, ≥12 paid conversions and ≥9 still active/renewing at the defined checkpoint; a failure threshold might be <40% paid conversion or <50% renewal among paid. These are classroom pre-commitments, not universal industry benchmarks.

Alternative strong answer: test whether acquisition remains viable when founder hours are constrained and a repeatable channel owns the motion.

9 — FOUNDER MEMO

WHY NOW: Restaurants in the sample show recurring missed-call pain and active manual workarounds while voice AI can now handle a meaningful share of routine calls.

STRONGEST EVIDENCE: Customers progressed beyond interviews to installation, payment and early renewal, with one second-outlet request.

BIGGEST UNKNOWN: Whether delivered value and retention remain strong enough at scale without founder-heavy selling/support.

NEXT 30-DAY BET: Run a paid, tightly defined ICP cohort that measures conversion, resolution quality, renewal and founder-hours per acquired account.

WHAT WOULD CHANGE/KILL: Materially weak paid conversion/renewal or inability to deliver acceptable multilingual/order accuracy without excessive human intervention would weaken the thesis.

WHAT YOU NEED NEXT: More paid cohort evidence, repeatable acquisition data and product evidence on the specific call types that drive restaurant value.

10 — THE CHEQUE

No single verdict is mandatory. A particularly defensible answer is 'Do not invest the full ₹75L unconditionally yet; require more evidence or milestone/tranche the capital,' because the case has genuine behavioural evidence but immature retention, founder-heavy acquisition and unresolved product/value contradictions.

Reasons supporting continued diligence/investment: observable pain/workarounds; payment/renewal; one expansion request; plausible narrow ICP.

Risks: tiny/young retention cohort; acquisition cost boundary excludes founder labour; 96% answer rate overstates autonomous resolution; multilingual/order performance; multi-city repeatability unproven.

Milestones should be numeric and hypothesis-linked, e.g. a larger paid cohort, defined renewal/retention threshold, founder-hours/customer reduction, resolution quality, or channel CAC under a consistent definition.

A kill condition should target the thesis, not vanity metrics.

Strong completion: 'The next ₹75L should buy evidence that the company can repeatedly acquire, deliver measurable value to, and retain a narrow ICP without founder heroics before geographic expansion.'