zap-coding-agent Tan — Startup Strategy & Company Building
The Only Question That Matters Early
"Do people want this? Not 'would people use this if it existed' — do people want it badly enough to change their behavior?" — Tan / YC axiom
Every other question is downstream of this one. Revenue model, scaling, team structure, fundraising — none of these matter if the answer is no. The entire early-stage strategy is designed to get a reliable answer as fast as possible.
Product-Market Fit — Recognizing It and Its Absence
PMF is not a survey result. Users who say "yes I'd pay for that" in an interview are not PMF. PMF is observable behavior:
| Signal | What it means |
|---|---|
| Users come back without being asked | Retention is organic — the product is becoming a habit |
| Users tell other users without incentive | Word-of-mouth NPS > 50 is a PMF indicator |
| Users are upset when the product is down | Dependency has formed |
| Users use the product in ways you didn't design for | Demand exceeds the spec — there's a deeper need |
| Revenue grows without a commensurate increase in sales effort | Pull, not push |
The Sean Ellis PMF test: Survey active users: "How would you feel if you could no longer use [product]?" If >40% say "very disappointed," you likely have PMF.
The absence of PMF looks like: high activation, low retention. Users sign up, try it once, and don't come back. No amount of growth hacking, better onboarding, or feature additions fixes a retention problem that comes from a missing core value.
The Pivot vs Persist Decision
This is the hardest judgment call in a startup. Tan's framework:
Persist when:
- Retention is good but growth is slow → distribution problem, not product
- Early users love it but the market hasn't heard of you → discovery problem
- You have strong conviction the market will develop (timing play)
- You haven't yet done enough user interviews to understand why retention fails
Pivot when:
- Retention is bad AND you've done the user research AND you understand why
- The users who are retaining are not the users you can build a business on
- You've been at it for 12+ months with no meaningful retention signal
- Your best users are using the product for a different problem than you built it for (→ pivot toward them)
The pivot trap: pivoting to escape discomfort without understanding why retention fails. You take the same lack of understanding to the new problem.
Fundraising Readiness — Tan's Criteria
"Investors invest in lines, not dots. You need to show a trajectory, not a state."
The criteria for raising a Seed:
- A clear founder thesis: why are you the right people to solve this problem? (insight, experience, or obsession)
- A believable market: is the problem real, is the market large enough, and do you have a plausible path to a significant share of it?
- Evidence of demand: usage, retention, revenue, letters of intent — in roughly that order of strength
- A good answer to "why now?": what has changed in the world that makes this possible/necessary now that it wasn't 5 years ago?
What NOT to have at Seed:
- A complete product (MVPs are expected)
- Profitability (not expected)
- A full team (expected to hire with the raise)
The pitch is a story, not a slide deck. Tan's formulation: "We noticed [specific problem observed directly]. We tried [existing solutions]. They fail because [specific mechanism]. We built [specific thing] that [specific outcome]. It worked: [evidence]. We need [$] to [specific milestone that de-risks the next round]."
Early Team Building — the Hiring Principles
Tan's framework for early hires (first 10):
Hire for the problem, not the role. "We need a VP Engineering" is a job description, not a problem to solve. "We need to ship 3× faster" is a problem — the hire should solve it.
Mission alignment over experience. An average engineer who cares deeply about your mission and will figure things out outperforms a better engineer who is mercenary.
The reference check is the interview. Ask references: "If you were starting a company, would you hire this person?" and "What would this person need to work on?" Listen to the hesitations.
Slow to hire, fast to fire. A bad early hire is existential. One misaligned person in a 5-person team is 20% of the culture.
The airport test revisited. Not "would I enjoy being stuck with this person" — but "would I trust this person to make a hard decision without me?" Early teams need low-supervision execution.
The YC Interview Question as a Product Filter
YC's interview questions are a forcing function for clarity. Apply them to any major product decision:
- "What have you built?" → Have you shipped, or are you planning?
- "Who is your user, specifically?" → "Everyone" is not an answer. Name a person.
- "What do users do differently because of you?" → Behavioral change, not features.
- "What's the hardest part of this?" → If you don't know the hardest part, you haven't started.
- "How does this grow?" → What's the growth mechanism, and is it inherent to the product?
- "What would make you a billion-dollar company?" → Separates feature companies from platform companies.
The discipline: answer these before your roadmap planning, not just before investor meetings.
The Compounding Advantage of Taste
Tan's singular view on what separates breakout companies: the founders have taste. They know what "great" looks like before they've built it. They can feel when something isn't there yet. They don't ship mediocre because they can't feel the difference between good and mediocre.
Taste is not a personality trait — it is built by exposure, study, and feedback loops. Engineers who invest in design taste (see tan-design) close the loop between their instinct and their output.