There are two ways to start a company. One is to start with an idea. The other is to start with a problem. The first one feels more heroic. The second one works more often.
The reason is that ideas are cheap in a very specific way: they can exist without touching reality. Problems can't. A problem has an owner. It happens at a certain time in a certain place. It has workarounds with names. If you try to talk about a problem for ten minutes without drifting into vagueness, you discover whether it's real.
So the first piece of advice is almost boring: be painfully specific. Don't build "a tool for creators." Build something that saves wedding photographers two hours per shoot, or helps accounting teams close books a day faster, or lets teachers grade essays without turning Sunday into a small tragedy.
People resist this because specificity feels like shrinking. But in startups, shrinking is how you aim. The wide market is a mirage from a distance. Up close, it's a thousand narrow markets. The ones you can win are the ones you can name.
Once you pick a problem, you have to do the second thing founders resist: talk to customers before you write code. This sounds like the kind of advice you'd put on a poster and then ignore. But it's not about being virtuous. It's about time. Code is slow, and it's a one-way door in the worst sense: once you've built a thing, you start protecting it with your ego. You don't just ship software, you ship excuses.
So you want to learn while it's still cheap to be wrong.
The trick is to have conversations that produce evidence, not applause. Don't ask "Would you use this?" People will say yes the way they say they should drink more water. Ask about the last time the problem happened. Ask what it cost. Ask what they tried. Ask what they do now, and what they hate about it. Then you listen for repetition. One person can be an interesting story. Ten people saying the same thing is a map.
When you do build, build a wedge. "Platform" is usually a synonym for "I don't know what the first user is doing on Tuesday." A wedge is the opposite. It's one use case that is clearly, almost embarrassingly better for a defined group. The best wedges feel like cheating. They work because you're willing to do something narrow and sharp, and other people are still describing an ecosystem.
A wedge also forces you to decide what you will not do. That's the part founders avoid, because not doing things feels like leaving money on the table. But early on there is no table. There is only time and attention. Your job is to spend them on the smallest thing that can become undeniably valuable.
This is why shipping matters so much. You want tiny releases, weekly if you can manage it, because they make you honest. Every week you either show users something new or you notice you didn't. Weekly shipping turns "someday" into "Friday." It also prevents your product from becoming a museum of imagined requirements.
Most people think startups win by doing more. Many actually win by doing less, but doing it relentlessly.
If you want one metric to keep you from lying to yourself, pick retention. Early growth is often decoration. It's surprisingly easy to get people to click a link once. It's harder to get them to come back when the novelty is gone and the product has to earn its place on their screen. A small number of users who would be annoyed if you disappeared is worth more than a large number who think your idea is neat.
To get there, you end up doing things that don't scale. This phrase gets repeated so much it's started to sound like a slogan, but it points to a real pattern. In the beginning you will do onboarding calls, manually set things up, fix edge cases by hand, write one-off scripts, customize what you swore would be standardized. It can feel wrong, like you're betraying the purity of software. But what you're really doing is buying understanding. You are finding the parts that matter enough to automate later.
The mistake is to confuse "not scalable" with "not worth doing." The most scalable thing you can do early is learn what people actually need.
There's another mistake that looks more sophisticated: ignoring distribution. Founders will treat marketing like a dirty word, as if building is the noble part and getting customers is the part you outsource to people who like LinkedIn. But distribution is not an accessory. It is a component. A product without a channel is a hobby with servers.
This doesn't mean you need a grand plan. It means you need a real first path: a community where your users already are, a niche newsletter, a marketplace, outbound to a very specific role, content that answers the exact questions they type into Google at 11pm. The best channel is the one you can execute with the resources you have and the patience you don't.
All of this is happening under a deadline you can't negotiate: runway. If you don't know your burn, you are driving at night without headlights. Many startups die for a very ordinary reason. They run out of time. It rarely looks dramatic from the inside. It looks like a few months of "We're close," followed by a few weeks of quiet panic.
So be serious about money earlier than you think you should. Keep fixed costs low. Preserve optionality. Raise when you can, not when you must. And if you're bootstrapping, be even more allergic to anything that turns into a monthly obligation.
At some point the conversation turns to cofounders. People talk about them like they're a cheat code. They can be. They can also be the single most expensive mistake you make. A bad cofounder fit doesn't just slow you down. It breaks decision-making, drains morale, and turns every hard moment into a referendum on the relationship.
The right test isn't "Are we friends?" It's "Can we do ugly work together, under stress, with incomplete information, and still trust each other tomorrow?" If the answer is yes, write down roles, equity, and how you break ties before you need to. If the answer is no, don't try to fix it with optimism.
Finally there's the advice nobody likes because it sounds like life coaching: protect your energy. The work is not just hours. It's judgment. You make a thousand small decisions, and each one costs a little. If you sleep badly, eat badly, never move, and mainline stress, your decision quality degrades until you're basically building by superstition.
Ambition is not an excuse to become unreliable.
The hopeful thing is that none of this requires genius. It mostly requires honesty and persistence. Talk to the people with the problem. Pick the narrow thing you can win. Ship small, often. Watch whether they come back. Do the unscalable work that teaches you what matters. Don't pretend distribution will happen later. Don't run out of time.
Starting a company is hard, but the path is not mysterious. It's just easy to walk with your eyes closed.