- Start small and build as little as possible. Most of what founders build early is never used, and every feature you ship is one you have to keep alive.
- Solve a real, urgent problem and make people come back to it. A painkiller people return to weekly beats a vitamin they admire once.
- You're building a business, not buying a lottery ticket. Play for a compounding position, not a day-one jackpot.
- If your product has network effects, design a one-player version that is useful before the network exists. A marketplace with two people in it is worthless.
- You cannot beat incumbents on breadth. Pick a narrow group, understand them better than anyone, and win on depth, service and care.
There is no shortage of startup advice, and most of it is useless because it is written to be admired rather than used. What follows is the opposite: ten rules I keep returning to with founders, most of them learnt the hard and expensive way, plus a bonus for anyone whose product depends on network effects. They are not a formula, because there isn't one. They are a set of defaults that keep you alive long enough to find the thing that works. Take them in order, because the order matters more than any single point.
01Keep it small.
Small team, small burn, small surface area. A startup's real advantage over an incumbent is not money or reach, it is speed and focus, and both of those die the moment you get big before you are ready. A founder plus one or two people who can actually build will out-manoeuvre a bloated team every time. Keep the headcount down, keep the monthly costs down, and keep the number of things you are trying to do down. Small is not a limitation you tolerate until you can afford more. Small is the weapon.
02Build as little as possible.
The instinct is to build the whole vision. Resist it. The goal at the start is not to ship a product, it is to learn whether anyone wants it, and you can learn that with a fraction of what you think you need. Most features in an early product are never touched, and every one you ship is something you then have to maintain, explain and defend. Build the smallest thing that tests the idea, put it in front of real people, and let what they do decide what you build next. If you can test the idea without writing code at all, do that first.
03Make a painkiller, not a vitamin.
A vitamin is nice to have. People nod, they say it's a lovely idea, and then they carry on exactly as before. A painkiller solves a problem that genuinely hurts, and people will find the money and the time for it because the pain of not having it is worse. Before you build anything, be honest about which one you are making. If your best description of the product is that it would be good for people, you have a vitamin. If people are already hacking together an ugly workaround because they need the problem solved today, you have a painkiller. Chase the pain.
04Play the long game, not the lottery.
You are building a business, not buying a lottery ticket. The version of startups sold on stage is a rocket to a billion-pound exit, and it quietly teaches founders to make decisions as if only the jackpot counts: raise huge, spend huge, grow at any cost, win on day one or die. Almost no one wins that way, and the all-or-nothing mindset burns the cash and the goodwill you needed for the long run. A real business compounds. It gets a bit better, a bit bigger and a bit more trusted every month, and it is still standing in three years when the lottery players have folded. Don't try to win on day one. Try to still be playing on day one thousand.
05Build a habit.
The single best signal that you have something real is that people come back on their own. Not because you emailed them, not because you ran an ad, but because the product has become part of how they do a thing. That is what traction actually is, and it is far more honest than sign-up numbers, which measure your marketing rather than your product. So design deliberately for the return: give people a reason to come back this week, and then next week, and make each return a little more valuable than the last. Retention is the number that tells the truth, and a growing base of people who keep coming back is the clearest proof you are onto something. If you want the mechanics of how habits form around a product, the Hooked model is a good place to start.
06Earn trust.
You are the new name in the room, and nobody owes you their data, their money or their workflow. Established competitors have spent years earning the benefit of the doubt that you have to earn from scratch. So treat trust as a product feature you build on purpose. Be clear about what you do and don't do, do the unglamorous things well, answer quickly when something breaks, and never over-promise to close a deal. Every kept promise is a deposit, and early on those deposits are the only capital you have that a bigger competitor can't simply outspend.
07Know exactly who you're building for.
Not a demographic, not a market the size of a country, but a specific kind of person with a specific problem you can describe in a sentence. The more precisely you can name who it's for, the sharper every decision downstream becomes: what to build, what to say, where to find them, what to charge. Vague targeting produces vague products that are nobody's first choice. If you cannot picture the exact person who will be delighted by this, you are not ready to build it yet.
08Don't build for everyone at once.
Trying to serve everyone is the fastest way to serve no one. The temptation is understandable, because a bigger addressable market looks like a bigger opportunity, but in practice it means a product pulled in ten directions, a message that speaks to no one in particular, and a roadmap that never goes deep enough to be loved. Pick one group and commit to them completely. You can always expand later from a position of strength. You cannot expand from a product that was mediocre for everybody.
09Serve a narrow beachhead ruthlessly.
Once you have chosen your one group, over-serve them. Understand them better than any competitor bothers to. Solve their specific problem so completely that switching to you feels obvious. This is the whole strategy, not a phase you rush through on the way to the mass market. A small group that loves you is worth more than a large group that mildly tolerates you, because the ones who love you stay, they tell others like them, and they give you the honest feedback that makes the product genuinely good. Depth first, breadth later.
10Win where you can.
Be honest with yourself: on most measures you are not as good as the incumbents. They have been at it longer, they have refined their product over years, and they have resources you don't. Accept that, because it is freeing. You do not need to be better at everything, you need to be better at the few things your narrow group cares about most, and those are usually things a big company is bad at. The niche problem they ignore because it is too small for them. The customer service they cannot be bothered to do well. The speed, the care, the feeling of being understood. Pick the ground where a large, slow competitor cannot easily follow, and win there completely. That is how small companies beat big ones. Not by fighting on their terms, but by refusing to.
BonusDesign a one-player start.
This one is for anyone building a product with network effects: a marketplace, a social product, a community, anything that gets better as more people join. Everyone loves network effects, and rightly so, because once they take hold they are the strongest moat there is. But they are a trap at the start, because a network product is worth nothing until the network exists. A marketplace with two people in it is not a small marketplace, it is nothing. A social app where none of your friends have joined is an empty room. Launch and wait for the flywheel to start turning on its own, and you run out of money before it does.
The way through is to design a version of the product that is genuinely useful to one person on their own, before anyone else turns up. Single-player value worth using even when the network is empty. That is what buys you the time to reach the density where the network effects finally take hold. It is a design problem you engineer your way through on purpose, not a budget problem you spend your way out of. I wrote a whole book on precisely this, Cold Start by Design, because it is the part of network products that founders get wrong more than any other.
The through-line.
Read the ten together, with the bonus alongside them, and they say one thing. Stay small so you can move, build little so you can learn, solve real pain so people care, make it a habit so they stay, and if it is a network product make it useful on day one before the network arrives, earn the trust you haven't got yet, and then pick a fight narrow enough that you can actually win it. None of it is about being the biggest or the loudest or the best-funded. It is about being the sharpest in a space small enough to dominate, for long enough that compounding does the rest. Do that, and you don't need to win the lottery. You just need to keep playing while everyone else folds.