A wiki about what actually creates durable competitive advantage in startups — and what the rest of us tell ourselves to feel better about not having any.
The core question
Most startups die. The interesting ones don’t die for the reasons they tell investors. They die because they spent two years executing a playbook that was never going to work — there was no Power at the end of it. Hamilton Helmer’s framework gives the cleanest articulation of what Power actually means: Benefit + Barrier. Benefit without Barrier is a feature. Barrier without Benefit is a moat around an empty castle. You need both, and there are only seven structural ways to get them. Each has a window in which it can be established (power-progression) — miss the window and the door closes for good.
Key sources
How it connects
┌─────────────────────┐
│ SEVEN POWERS │
│ (the framework) │
└────────┬────────────┘
│
┌──────────────┼──────────────┐
▼ ▼ ▼
┌────────────┐ ┌────────────┐ ┌────────────┐
│ MONOPOLY │ │ COUNTER- │ │ NETWORK │
│ vs. │ │ POSITIONING│ │ EFFECTS │
│COMPETITION │ │ │ │ │
└──────┬─────┘ └──────┬─────┘ └──────┬─────┘
│ │ │
▼ │ ▼
┌────────────┐ │ ┌────────────┐
│ CROSSING │◄────────┘ │ ATOMIC │
│ THE CHASM │ │ NETWORK │
└──────┬─────┘ └──────┬─────┘
│ │
└───────────┬───────────────────┘
▼
┌────────────────┐
│ GROWTH AS │
│ COMPASS │
└───────┬────────┘
│
┌────────────┼────────────┐
▼ ▼ ▼
┌───────────┐ ┌───────────┐ ┌───────────┐
│ HOOK │ │DISTRIBUTION│ │ PRODUCT │
│ MODEL │ │ │ │ NARRATIVE │
└───────────┘ └───────────┘ └───────────┘
Start at the top: seven-powers defines what durable advantage looks like. Move down: monopoly-vs-competition, counter-positioning, and network-effects are the primary strategies for building it. Cross the chasm: both crossing-the-chasm-concept and atomic-network solve the same problem — going from early traction to real market presence. Grow: growth-as-compass ties it together. Execute: hook-model, distribution, and product-narrative are how you build the engine.
The landscape
The seven powers. seven-powers is the backbone. Each Power type — scale-economies, network-effects, counter-positioning, switching-costs, branding, cornered-resource, process-power — has a specific Benefit/Barrier structure and a window in which it can be established (power-progression). Paul Graham’s startup philosophy maps onto this framework with surprising precision: growth opens the windows, technical taste creates cornered resources, and fighting in new markets is counter-positioning by another name (graham-on-power).
Monopoly and competition. Thiel’s monopoly-vs-competition is the thesis: competition destroys profits, build a monopoly instead. Start small, dominate a niche, then expand (sequencing-markets) — a pattern echoed by Moore’s crossing-the-chasm-concept and Chen’s atomic-network. Three different frameworks, one playbook. The frameworks differ in vocabulary but agree on the move: pick a fight you can win and own it before anyone notices.
Product-market fit. Andreessen’s foundational claim: the only thing that matters for a startup is getting to PMF, and the dominant variable is market, not team or product (only-thing-that-matters). For decades that was a vivid feeling — “you can feel it when it’s happening” — but not a measurement. Rahul Vohra changed that in 2018 with the Sean Ellis 40% test, turning PMF into a number you can move week over week (superhuman-pmf-engine). The cleanest real-world demonstration of all of this is ChatGPT — the fastest consumer adoption in history, accidental, undeniable, and a perfect specimen of “the market pulled the product out of the startup.”
Growth and moats. growth-as-compass — growth rate is the measure of a startup and the compass for every decision. But growth has invisible ceilings (invisible-asymptotes), and moats are moving targets as AI collapses development costs (software-dev-costs-moats) and traditional switching costs erode. Retention drives growth more than acquisition (duolingo-growth, pmf-roadmap). Growth loops compound where funnels don’t (growth-loops).
AI and the new moat landscape. The AI era is reshaping what’s defensible. Open-source models erode proprietary AI advantages (google-no-moat). Data moats — proprietary data populating the context window — emerge as the new cornered-resource (cheating-is-all-you-need, data-moat). Startups win in AI where workflow tools are weak and human-in-the-loop is natural (ai-startup-vs-incumbent). When anyone can build software, distribution becomes the primary moat. ChatGPT is the proof of concept on every one of these claims at once (chatgpt-pmf).
Operational strategy. Two philosophies, both effective. Amazon’s working-backwards-method: PR/FAQ documents, single-threaded-leadership, two-pizza-teams, mechanisms over intentions. Boyd’s ooda-loop: fluid maneuver through tempo and asymmetric-fast-transients, direction-setting over goal-setting. Netflix offers a third path: replace process with talent density (netflix-culture, keith-rabois-lessons).
Product and distribution. simplicity-as-strategy — simplicity is a competitive advantage that erodes over time. hook-model — habits form through trigger-action-reward-investment loops. distribution matters more than building when dev costs approach zero. product-narrative — your startup is a movement, not a company. grand-slam-offers — make offers so good they sell in a category of one.
Navigation. idea-maze-concept — good founders navigate a maze of possible paths, not a single idea. product-market-fit is necessary but not sufficient — Gumroad had PMF but not market fit (gumroad-failure). The idea-maze-concept is how you avoid spending years on the wrong path. Products succeed when their atomic-concepts match how customers think about their workflow.
See also: definite-optimism (Thiel), toyota-production-system (Boyd + Helmer), aggregation-theory (Thompson), single-user-utility (bootstrap for networks).
Current state
Seeded from Readwise on 2026-04-10. 52 source pages, 43 concept pages, 95 total. Most recent addition: a case study on ChatGPT’s path to PMF — the cleanest specimen of “the market pulls the product out of the startup” we are likely to get. See index for the full catalog.