Answer 1
In terms of when to stop and pivot: PART II — THE COLD START PROBLEM
**Ch. 4 — Tiny Speck / Slack.** Tiny Speck spent 4 yrs 10 mo, raised $17M, hired 45 people on the multiplayer game **Glitch** — Butterfield: "97% who signed up would be out of there within five minutes" (leaky bucket). Relaunched as **Slack** → 20M DAU, ~1M businesses, exited to Salesforce for ~$26B, $800M+ revenue. Slack grew from an internal IRC-based "frankentool" (Slack = Searchable Log of All Conversation and Knowledge). **[BIZBUILDER]** Butterfield personally signed up 45 companies in private beta — "I just had friends at other companies" — and personally handled the …
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 2
In terms of when to stop and pivot: Quick-reference — named tactics and when they work/fail
| Tactic | Works when | Fails when | |---|---|---| | **Atomic network** | Pick the tiniest specific group at a specific time; build density | "Peanut-buttering" across a whole geography/industry | | **Solve a Hard Problem** | Product nails the hard side's unaddressed need (Tinder for women) | Hard side churns → degrades for everyone | | **Come for the Tool, Stay for the Network** | Tool + network tightly integrated (Dropbox folders) | Tool/network divergent → low conversion | | **Invite-Only** | Curated connected users invite connected users | Used purely for hype; or kills…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 3
In terms of when to stop and pivot: Central framework: Cold Start Theory
Network effects are not binary — they are a **lifecycle of five sequential stages**, each with its own goal and tactics: 1. **The Cold Start Problem** — launch; no users; anti-network effects dominate; most networks die here. 2. **Tipping Point** — a *repeatable* strategy to launch network after network; each tips faster, like dominoes. 3. **Escape Velocity** — furiously strengthening the trio of network forces to sustain rapid growth at scale. 4. **Hitting the Ceiling** — growth stalls; saturation, CAC spikes, fraud, overcrowding, context collapse. 5. **The Moat** — using …
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 4
In terms of when to stop and pivot: PART I — NETWORK EFFECTS
**Ch. 1 — What's a Network Effect, Anyway?** A network effect = product gets more valuable as more people use it. It has a **duality**: product (software) + network (people). Theodore Vail (AT&T, 1900): "A telephone without a connection at the other end of the line... is one of the most useless things in the world." 1908: <5M phones for ~90M Americans. The "Billion Users Club": leading social network 2B+ DAU; YouTube ~2B users; Apple 1.6B iOS devices; Google 3B; Facebook 2.85B; Microsoft 1.5B Windows + 1B Office. Network ≠ ownership (Airbnb owns no rooms, Apple owns no apps…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 5
In terms of when to stop and pivot: PART VI — THE MOAT
**Ch. 29 — Wimdu versus Airbnb.** If your product has network effects, your competitors likely do too. **Wimdu** — a near-exact Airbnb clone from the Samwer brothers' Rocket Internet (2011), launched with $90M funding, 400+ employees, "ten times bigger than Airbnb on paper." Airbnb was then 2.5 yrs old, 40 employees, USD-only. Wimdu scraped Airbnb listings, posed as guests to recruit Airbnb hosts, built 50,000+ listings — then **went to zero** by 2014–2018. **"All supply isn't created equal"** (Airbnb employee #17): "Wimdu's top 10% of inventory was at the bottom 10% of Air…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md
Answer 6
In terms of when to stop and pivot: 5. Benchmark Reference Table
| Metric | Benchmark | Source | |--------|-----------|--------| | Hook rate (ideal) | 50% | Artemy, DDM conf | | Hook rate (acceptable) | 22-26% | Artemy, DDM conf | | Hook rate (kill threshold) | <15% | Artemy, DDM conf | | Video retention | 3.2-3.6 | Artemy, DDM conf | | Kling rework CPA improvement | ~20% | Artemy, DDM conf | | Kling rework volume needed | 20-30 variations per winner | Artemy, DDM conf | | Typical waste per missed kill signal | $3-16K per creative | Kirill, DDM conf | | Kill signal fires at | $100-$3K spend | Kirill, DDM conf | | Sasha subscription price…
Source: src/lib/bricks/sources/perf-marketing-playbook.md
Answer 7
In terms of when to stop and pivot: PART III — THE TIPPING POINT
**Ch. 11 — Tinder (Tipping Point).** The Tipping Point = a **repeatable strategy** to launch network after network. Tinder: 2B+ swipes/day, 1M dates/week, $1B+ revenue. Dating has naturally high churn (happy couples leave). **[SOCIAL][BIZBUILDER] The USC party tactic:** the team threw an incredible birthday party for a hyperconnected friend; to get in you had to download the Tinder app (bouncer checked) — highest one-day download spike, but what mattered was it being "**500 of the right people**" — the most social, hyperconnected people, on Tinder at the same time. **95% of…
Source: src/lib/bricks/sources/andrew-chen-cold-start-problem.md