TL;DR: Growth rate is the defining measure of a startup. If you get growth, almost everything else falls into place. If you don’t, almost nothing else can save you. Pick a target weekly growth rate, treat missing it as an emergency, and use it as a compass for every decision you make.

What it means

Paul Graham distills the startup to its essence: a company designed to grow fast. The growth rate — not revenue, not users, not features, not press — is the metric that matters. The math is unforgiving and worth burning into your retina (startup-equals-growth):

  • 1% weekly growth = 1.7× per year
  • 5% weekly = 12.6× per year
  • 10% weekly = 142× per year (and almost certainly unsustainable)

Small differences in weekly rate produce enormous differences over a year. The discipline is to pick a target weekly rate, try to hit it every week, and treat missing it as a failure that demands immediate investigation. This turns startup-building into a single optimization problem, and Graham argues that narrow focus is “wonderfully effective” — it cuts through the dozens of distractions a founder could be obsessing over and replaces them with one number that you either hit or you don’t.

The argument

Growth solves almost everything. Hiring, fundraising, morale, strategy — growth is the solvent. Sam Altman puts it bluntly: “Compounding success solves almost all internal problems, particularly hard ones” (hard-startups). The inverse is also true — without growth, every problem metastasizes. Hiring becomes harder because no one wants to join a flat company. Fundraising becomes harder because investors price startups on slope, not absolute size. Morale becomes harder because flat startups feel dead even when they’re not.

But growth has invisible ceilings. Every growth curve eventually slams into an asymptote you can’t see from inside the company (invisible-asymptotes). Amazon’s was shipping fees. Finding the asymptote early and routing around it is the difference between Amazon Prime and a stalled e-commerce business. The diagnostic question isn’t “how do we grow?” — it’s “why doesn’t every person in the world use our product yet?” The honest answer to that question, asked of the people who don’t convert, is the most valuable strategic input you can collect.

Growth creates windows for Power. The power-progression maps directly onto growth phases. During origination (pre-takeoff), you can lock in counter-positioning and Cornered Resources. During takeoff (rapid growth), you can establish Scale Economies, Network Economies, and Switching Costs. After takeoff, only Branding and Process Power are still available, and both take years. Miss your window and the door closes for good. Growth isn’t just a metric. It’s the engine that opens strategic possibilities.

Hard startups grow easier. This is the most counterintuitive piece of growth advice and the one founders most reliably reject when they first hear it. Working on a hard, significant problem generates a tailwind — talent, advice, partnerships, and funding flow toward ambition. “An easy startup is a headwind; a hard startup is a tailwind” (hard-startups). The pitch deck for a clone-of-X is a slog. The pitch deck for “we are going to fix Y, an enormous problem nobody has cracked” gets meetings, hires, and capital you would never get for the easier-looking version.

Revenue is the best measure. Active users is the next best. Graham notes that revenue grows roughly proportional to users for most early-stage startups, so user growth is a reasonable proxy when you’re not yet charging. But absolute numbers are meaningless without the ratio. Getting 100 new customers a month sounds good until you realize your growth rate is decreasing 0.5% per week. The ratio is the entire game.

The cleanest demonstration

ChatGPT is what growth-as-compass looks like at extreme scale. OpenAI didn’t have a “growth strategy” in November 2022 — they had a research preview. The growth happened to them. The reason it worked is that all of OpenAI’s downstream decisions (more compute, more hiring, more model capability, faster shipping cadence) were dictated by the growth rate they were trying to keep up with, not the other way around. Growth was the compass. Everything else was an output.

The rare lesson there is that organic growth-as-compass is the strongest signal there is. Bought growth doesn’t open windows, because the market isn’t actually pulling. Earned growth opens every window at once.

Loose threads

  • The growth-as-compass discipline is hardest to follow precisely when it matters most: in the messy middle, when you have some growth but not enough, and every week you’re tempted to chase a different shiny optimization.
  • Growth metrics can be gamed. Sustainable growth requires retention behind it (pmf-roadmap, duolingo-growth). A growth rate built on weak retention is a leaky bucket that looks fine in monthly numbers and devastating in cohort analysis.
  • See invisible-asymptotes for the other side: when growth is your compass, you have to know when the compass is broken because you’ve hit a ceiling that won’t move.

What links here