Startup Playbook

Startup Playbook. An essay by Sam Altman, from 2015.

Something users love

Altman's playbook starts with one goal: make something users love. Only then do you work out how to reach many more of them.

Today's giants all began with early users who loved the product so much they told their friends. Skip that step, and you fail.

So a few people who love it beats a crowd who merely like it. Adding users is the easy part. Turning like into love is the hard one.

Hard, but not risky

A fair warning: starting a startup is miserable. Y Combinator's founders keep saying it was far harder than they pictured.

But it isn't very risky for your career. If you're good at technology and it fails, there'll be jobs waiting.

In Altman's view, the bigger risk is the opposite: an idea you care about, and a safe, dull job where it never gets tried.

Success takes four things: the right idea, a strong team, an excellent product, and relentless execution. The playbook covers each in turn.

Questions about the idea

Y Combinator first asks what you are building, and why. A short, clear answer shows clear thinking; a tangled one is a warning.

Then: who badly needs this? Ideally, you are the user yourself. Next best, you know that user extremely well.

To test it: for consumer products, launch and watch. For businesses, try to sell it first, ideally getting a letter of intent before any code.

They ask why you care enough to last for years, and how it could become a monopoly: stronger as it grows, and hard to copy.

And the market. Oddly, the best answer is often a big share of a small market, one about to grow fast as technology shifts.

Ideas that sound bad

Y Combinator favours the new over the derivative, where ten times better counts as new. Ten teams cloning something that already exists? Skeptical.

Strangely, new and hard is easier than copied and easy. People want to help with, and join, the new thing.

The best ideas sound bad, but are good. So skip the secrecy; telling people brings help. Expect plenty to call it terrible, and grow a thick skin.

No idea yet? YC once funded teams without one. All of them failed, settling for plausible copies. Better to learn widely, notice problems, and let ideas come.

A great team

Part two: the team. Great companies need great founders: unstoppable, determined, formidable, resourceful. Experience matters far less.

The best ones answer fast and are easy to talk to. They're firm about the core mission, yet flexible about almost everything else.

Choose a cofounder you already know well, not a stranger from a matchmaking event. When prospects dip below zero, old loyalty keeps you both going.

Best: a good cofounder. Next: going solo. Worst by far: a bad cofounder. And settle the equity split early, keeping it close to equal.

A great product

Part three: the one thing every great company shares is a great product. Founders keep hunting for tricks, but a startup is where tricks stop working.

Growth hacks run out eventually. Over the long run, a company grows only because people want what it makes.

So build a product improvement engine: talk to users, watch them, find the weak spots, fix them, repeat. Five percent better each week compounds.

Get very close to your users. Let no one stand between them and the founders, who should handle sales and support themselves.

Do things that don't scale

Do things that don't scale. First users come one by one: Pinterest's Ben Silbermann approached strangers in Palo Alto coffee shops.

A big press announcement almost never does that. Instead, ship small pieces: start simple, and launch sooner than feels comfortable.

Struggling? Ask: do users come back? Are they fanatical? Would they be truly upset if you vanished? Do they recommend you unasked?

If not, the product usually just isn't good enough. Go talk to your users. And remember, support and sales are part of the product too.

Growth and momentum

Part four, execution. You still have to build the company yourself. Hoping a seasoned manager will do it for you is a famous dead end.

And growth fixes nearly everything. A growing company feels like it's winning. A stalled one fights over blame, and people burn out and leave.

So protect momentum. Pick one metric and make it the priority. Airbnb's founders put their target growth chart everywhere, even on the bathroom mirror.

Share the numbers openly, and beware vanity metrics: signups mean little without retention.

Growth traps

Growing fast while everything's held together with tape? That's good news: each fix buys more growth.

Plan for ten times your current size, not for massive scale. And give great customer service, even though it won't scale.

And beware the big press launch and partnership deals: they eat time and almost never work. Recruit users by hand, then try many channels and keep what works.

Focus and intensity

Asked to boil his advice on running a startup down to two words, Altman picks focus and intensity.

Focus means saying no a lot. Win at one thing before starting the next. Doing too many wrong things kills many startups.

Intensity means moving and deciding fast, on the right things. Paul Buchheit's rule: get 90% of the value from 10% of the effort.

Altman has never seen a slow founder truly succeed. And when early success brings panels and profiles, keep doing what got you there.

Jobs of the CEO

The CEO's one universal job: make sure the company wins. In practice: set the vision, evangelize, hire, raise money, and set the quality bar.

It takes over your life. There's room for one other big thing, like family. The highs and lows are intense and it's lonely, so find fellow CEOs to call.

It's a decade, not an all-nighter: eat, sleep, exercise. And make no excuses. Feel the unfairness for a minute, then go find the fix.

Sell everyone else a bold vision, but stay paranoid yourself. Be persistent, be optimistic, and write down the mission and values early.

One more trap: don't reinvent HR or finance. Save your creativity for the product, and copy what works elsewhere.

Hiring and managing

Altman's first advice on hiring: don't. The most successful companies waited. Employees cost money, add complexity, and make it harder to change course.

The best people want to join something that's clearly winning. Once you are, recruit hard: about a quarter of your time, with generous equity and trust.

Never lower the bar: if in doubt, the answer is no. Prefer aptitude to experience, and avoid people who are chronically negative.

Learn to manage, and resist hero mode, doing everything yourself until your team can't reach you. Keep everyone in one office, and fire quickly.

Competitors

New founders fear competitors. Altman's line: “99% of startups die from suicide, not murder.” Fix your own problems first.

So ignore competitors until their shipped product is actually beating yours. A press release is easier than code, and code is easier than a great product.

Every giant faced worse threats when it was small, and came through. There's always a counter-move.

Making money

Oh, and you have to make money: people must pay you more than it costs to deliver. Founders forget the cost part surprisingly often.

A free product can't buy its users; make something people share. If each customer is worth under $500 over their lifetime, you can't afford salespeople.

Below that line, try cheaper channels and earn back each customer's cost within three months. Above it, sell directly, starting with yourself.

Aim for ramen profitability: enough to feed the founders. Then investors no longer control your fate. And watch your cash obsessively.

Fundraising

Raise when you truly need money, or good terms are on offer. Too little can hurt, but too much almost always does: frugality slips.

The real secret is a good company. Investors want a shot at something huge; a sure ten-million-dollar outcome won't tempt them, even cheap.

When investors pass, trust the no, not the explanation. Anything short of a yes is a no, however hopeful it sounds.

Talk to investors in parallel, not one at a time; fear of missing out moves them. Keep terms clean, and don't fixate on valuation.

Landing the first check is hardest. Then the bar rises every round: a slick pitch that won a seed round won't carry a Series A.

In short

Make something a few users love before chasing many.

Stay close to users, and keep improving the product.

Guard momentum with focus and intensity.

Most startups fail from within: ignore the noise and execute.

The original is long, practical, and better than this. Read it: playbook.samaltman.com