hraness

saved

Making Startups Powerful

by Paul GrahamPaul Grahampublished

Hraness cites a source capture. The source author remains the source.

gist

Paul Graham’s office-hours heuristic for startups is not “how do we make more money?” but “what would make this company more powerful?” Owning the customer, routing money through you, inducing network effects or marketplaces, going full stack, catching tails that wag the dog, playing the long game, generosity and APIs, and selling early to fast-deciding customers all compound power; mafia markets and component boxes are what you escape by redesigning the idea.

ideas

  • Ask power, not only revenue. Revenue heuristics yield increments; power questions can reframe the company by orders of magnitude.
  • Own the outer surface. Prefer customer relationship, money flowing through you, app-store or marketplace shapes, and network effects—even induced ones like shared benchmarks or opt-in model training.
  • Transform the idea physically. Full-stack (eat the customer), hand-puppet outsourcing of brainwork, and peripheral features that become the real product (PayPal) are structural moves, not pricing tweaks.
  • Play the long game upstream. Cheap early users, generosity and open extensibility/APIs, early-stage customers who decide fast, and early data (Rippling) beat quarterly squeezing and enterprise mud.
  • Escape being held back. Component boxes and mafia markets (labels, PBMs) are defeated sideways by owning customers or making incumbents irrelevant, not by frontal product fights.

quotes

what would make this company more powerful?

Paul Graham, stating the governing office-hours heuristic.

It's always good when money flows through you.

Paul Graham, on owning the payment path.

you should create more value than you capture.

Paul Graham, citing Tim O'Reilly on generosity as power.

Upstream is almost always good, whether it's with money or user relationship or customer stage or data.

Paul Graham, summarizing the upstream pattern across money, users, stage, and data.